FAY, Circuit Judge:

In 1989, the Occupational Safety and Health Administration ("OSHA"), a division of the Department of Labor, issued its Air Contaminants Standard, a set of permissible exposure limits for 428 toxic substances. Air Contaminants Standard, 54 Fed.Reg. 2332 (1989) (codified at 29 C.F.R. § 1910.1000). In these consolidated appeals, petitioners representing various affected industries and the American Federation of Labor and Congress of Industrial Organizations ("AFL-CIO" or "the union") challenge both the procedure used by OSHA to generate this multi-substance standard and OSHA's findings on numerous specific substances included in the new standard. For the reasons that follow, we VACATE the Air Contaminants Standard and REMAND to the agency.

I. BACKGROUND

The Occupational Safety and Health Act of 1970 ("OSH Act" or "the Act"), 29 U.S.C. § 651-71, was adopted "to assure so far as possible every working man and woman in the Nation safe and healthful working conditions." Id. § 651(b). To this end, the Act authorizes the Secretary to issue occupational health and safety standards, id. § 655, with which each employer must comply. Id. § 654. Section 6(a) of the Act provided that in its first two years, OSHA should promulgate "start-up" standards, on an expedited basis and without public hearing or comment, based on "national consensus" or "established Federal standard[s]" that improve employee safety or health. Id. § 655(a). Pursuant to that authority, OSHA in 1971 promulgated approximately 425 permissible exposure limits ("PELs") for air contaminants, 29 C.F.R. § 1910.1000 (1971), derived principally from federal standards applicable to government contractors under the Walsh-Healey Act, 41 U.S.C. § 35. Air Contaminants Proposed Rule, 53 Fed.Reg. 20960, 20962 (1988).

The Act then provides two mechanisms to update these standards. Most new standards or revised existing standards must be promulgated under the requirements of section 6(b) of the OSH Act. 29 U.S.C. § 655(b). This section sets forth both procedural requirements and substantive criteria which the standards must meet. In promulgating these standards, OSHA must follow a procedure that is even more stringent than that in the federal Administrative Procedure Act, 5 U.S.C. § 553. United Steelworkers v. Marshall, 647 F.2d 1189, 1207 (D.C. Cir. 1980), cert. denied, 453 U.S. 913, 101 S.Ct. 3149, 69 L.Ed.2d 997 (1981). OSHA must provide notice of proposed rulemaking, give interested parties an opportunity to comment, and hold a public hearing if requested. 29 U.S.C. § 655(b)(2)-(4). As of 1988, OSHA had issued only twenty-four substance-specific and three "generic" health standards under section 6(b).

OSHA may also issue Emergency Temporary Standards under section 6(c) of the OSH Act, 29 U.S.C. § 655(c), when it determines "that employees are exposed to grave danger from exposure" to toxic substances. Id. § 655(c)(1). However, once OSHA has published an emergency standard, proceedings must commence for issuance of a regular standard under section 6(b). Id. § 655(c)(3).

On June 7, 1988, OSHA published a Notice of Proposed Rulemaking for its Air Contaminants Standard. 53 Fed.Reg. 20960-21393. In this single rulemaking, OSHA proposed to issue new or revised PELs for over 400 substances. OSHA limited the scope of this rulemaking to those substances for which the ACGIH recommended limits that were either new or more protective than the existing PELs. Id. at 20967. There was an initial comment period of forty-seven days, followed by a thirteen-day public hearing. Interested parties then had until October 7, 1988 to submit post-hearing evidence and until October 31, 1988 to submit post-hearing briefs.

OSHA then issued its revised Air Contaminants Standard for 428 toxic substances on January 19, 1989. 54 Fed.Reg. 2332-2983. This standard, which differs from the proposal in several respects, lowered the PELs for 212 substances, set new PELs for 164 previously unregulated substances, and left unchanged PELs for 52 substances for which lower limits had originally been proposed. Id. at 2334. The standard established an approximately four-year period for employers to come into compliance with the new standard using engineering and work practice controls. Id. at 2916. Until that time, employers may use respirators or any other reasonable methods to comply with the standards. Id. at 2915-16.

Various industry groups, the AFL-CIO, and specific individual companies filed challenges to the final standard in numerous United States Courts of Appeals. Pursuant to 28 U.S.C. § 2112(a), all petitions for review of the Air Contaminants Standard were transferred to this court, where they have been consolidated for disposition.

II. STANDARD OF REVIEW

Section 6(f) provides in relevant part that "the determinations of the Secretary shall be conclusive if supported by substantial evidence in the record considered as a whole." 29 U.S.C. § 655(f) (emphasis added). "Substantial evidence" is "`such relevant evidence as a reasonable mind might accept as adequate to support a conclusion.'" American Textile Mfrs. Inst., Inc. v. Donovan, 452 U.S. 490, 522, 101 S.Ct. 2478, 2497, 69 L.Ed.2d 185 (1981) (hereinafter "ATMI") (quoting Universal Camera Corp. v. NLRB, 340 U.S. 474, 477, 71 S.Ct. 456, 459, 95 L.Ed. 456 (1951)). Under this test, "we must take a `harder look' at OSHA's action than we would if we were reviewing the action under the more deferential arbitrary and capricious standard applicable to agencies governed by the Administrative Procedure Act." Asbestos Info. Ass'n v. OSHA, 727 F.2d 415, 421 (5th Cir. 1984) (footnote omitted); see also National Grain Feed Ass'n v. OSHA, 866 F.2d 717, 728 (5th Cir. 1988). Considering the record "as a whole" further requires that reviewing courts "take into account not just evidence that supports the agency's decision, but also countervailing evidence. . . . Yet this requirement does not alter the court's fundamental duty to uphold the agency's `choice between two fairly conflicting views, even though the court would justifiably have made a different choice had the matter been before it de novo.'" AFL-CIO v. Marshall, 617 F.2d 636, 649 n. 44 (D.C. Cir. 1979) (quoting Universal Camera Corp., 340 U.S. at 488, 71 S.Ct. at 464-65), aff'd in relevant part, ATMI, 452 U.S. 490, 101 S.Ct. 2478, 69 L.Ed.2d 185 (1981).

The substantial evidence test applies to review of policy decisions as well as factual determinations, see Texas Indep. Ginners Ass'n v. Marshall, 630 F.2d 398, 404 (5th Cir. 1980), even though policy decisions are "'not so susceptible to verification or refutation by the record.'" Id. (quoting American Petroleum Inst. v. OSHA, 581 F.2d 493, 497 (5th Cir. 1978), aff'd sub nom., Industrial Union Dept., AFL-CIO v. American Petroleum Inst., 448 U.S. 607, 100 S.Ct. 2844, 65 L.Ed.2d 1010 (1980). "[I]n setting an element of the standard OSHA must demonstrate substantial evidence for all matters of determinable fact and, on matters having no possible basis in determinable fact, must explain the relevant considerations on which it relied and its reasons for rejecting alternate views." United Steelworkers, 647 F.2d at 1253. OSHA's policy decisions must be: (1) consistent with the language and purpose of the OSH Act, and (2) reasonable under the rulemaking record. Texas Indep. Ginners, 630 F.2d at 404; see also National Grain Feed, 866 F.2d at 729.

Furthermore, "the validity of an agency's determination must be judged on the basis of the agency's stated reasons for making that determination." Industrial Union Dept., AFL-CIO v. American Petroleum Inst., 448 U.S. 607, 631 n. 31, 100 S.Ct. 2844, 2858 n. 31, 65 L.Ed.2d 1010 (1980) (plurality opinion) (hereinafter "Benzene"). Section 6(e) of the OSH Act provides that "[w]henever the Secretary promulgates any standard . . . he shall include a statement of the reasons for such action, which shall be published in the Federal Register." 29 U.S.C. § 655(e). In that statement of reasons,

the agency must pinpoint the factual evidence and the policy considerations upon which it relied. This requires explication of the assumptions underlying predictions or extrapolations, and of the basis for its resolution of conflicts and ambiguities. In enforcing these requirements, the court does not reach out to resolve controversies over technical data. Instead it seeks to ensure public accountability.

AFL-CIO v. Marshall, 617 F.2d at 651 (footnotes omitted). "[T]he courts will not be expected to scrutinize the record to uncover and formulate a rationale explaining an action, when the agency in the first instance has failed to articulate such rationale." ATMI, 452 U.S. at 539 n. 73, 101 S.Ct. at 2506.

III. DISCUSSION

In challenging the procedure by which OSHA promulgated the Air Contaminants Standard, a group of industry petitioners complain that OSHA's use of generic findings, the lumping together of so many substances in one rulemaking, and the short time provided for comment by interested parties, combine to create a record inadequate to support this massive new set of PELs. The union also challenges the rulemaking procedure utilized by OSHA for the Air Contaminants Standard. Not surprisingly, however, the union claims that this procedure resulted in standards that are systematically underprotective of employee health. The union further challenges OSHA's decision to limit the scope of the rulemaking to substances for which the ACGIH recommendation was more protective than the current PEL, and thereby to ignore both other air contaminant substances in need of regulation and standards for exposure monitoring and medical surveillance. Moreover, the union argues that there is no record support for a four-year compliance period for these standards, given that OSHA itself found that the standards can be met by existing technology.

A. "GENERIC" RULEMAKING

Unlike most of the OSHA standards previously reviewed by the courts, the Air Contaminants Standard regulates not a single toxic substance, but 428 different substances. The agency explained its decision to issue such an omnibus standard in its Notice of Proposed Rulemaking:

OSHA has issued only 24 substance-specific health regulations since its creation. It has not been able to review the many thousands of currently unregulated chemicals in the workplace nor to keep up with reviewing the several thousand new chemicals introduced since its creation. It has not been able to fully review the literature to determine if lower limits are needed for many of the approximately 400 substances it now regulates.

Using past approaches and practices, OSHA could continue to regulate a small number of the high priority substances and those of greatest public interest. However, it would take decades to review currently used chemicals and OSHA would never be able to keep up with the many chemicals which will be newly introduced in the future.

53 Fed.Reg. at 20963. For this reason, "OSHA determined that it was necessary to modify this approach through the use of generic rulemaking, which would simultaneously cover many substances." 54 Fed. Reg. at 2333 (emphasis added).

"Generic" means something "common to or characteristic of a whole group or class; typifying or subsuming; not specific or individual." Webster's Third New International Dictionary 945 (1966). Previous "generic" rulemakings by OSHA have all dealt with requirements that, once promulgated, could be applied to numerous different situations. For example, OSHA's Hazard Communication Standard, 29 C.F.R. § 1910.1200, mandates that employers inform employees of potentially hazardous materials. The regulation includes a basic list of substances which employers must treat as hazardous, but requires that the employers themselves also evaluate substances produced in their workplaces to determine if they are potentially hazardous based on available scientific evidence. United Steelworkers v. Auchter, 763 F.2d 728, 732 (3d Cir. 1985). Similarly, OSHA has issued standards regulating employee access to medical and toxic substance exposure records, 29 C.F.R. § 1910.20, and setting forth uniform criteria for application in future regulation of exposure to carcinogens, 29 C.F.R. Part 1990.

By contrast, the new Air Contaminants Standard is an amalgamation of 428 unrelated substance exposure limits. There is little common to this group of diverse substances except the fact that OSHA considers them toxic and in need of regulation. In fact, this rulemaking is the antithesis of a "generic" rulemaking; it is a set of 428 specific and individual substance exposure limits. Therefore, OSHA's characterization of this as a "generic" rulemaking is somewhat misleading.

Nonetheless, we find nothing in the OSH Act that would prevent OSHA from addressing multiple substances in a single rulemaking. Moreover, because the statute leaves this point open and because OSHA's interpretation of the statute is reasonable, it is appropriate for us to defer to OSHA's interpretation. See Chevron U.S.A., Inc. v. Natural Resources Defense Council, Inc., 467 U.S. 837, 842-45, 864-66, 104 S.Ct. 2778, 2781-83, 2792-93, 81 L.Ed.2d 694 (1984); see also Vermont Yankee Nuclear Power Corp. v. Natural Resources Defense Council, Inc., 435 U.S. 519, 524-25, 543-49, 98 S.Ct. 1197, 1211-15, 55 L.Ed.2d 460 (1978) (Courts should not impose procedural requirements on agencies if not statutorily required.). However, we believe the PEL for each substance must be able to stand independently, i.e., that each PEL must be supported by substantial evidence in the record considered as a whole and accompanied by adequate explanation. OSHA may not, by using such multi-substance rulemaking, ignore the requirements of the OSH Act. Both the industry petitioners and the union argue that such disregard was what in essence occurred. Regretfully, we agree.

B. SIGNIFICANT RISK OF MATERIAL HEALTH IMPAIRMENT

Section 3(8) of the OSH Act defines "occupational health and safety standard" as "a standard which requires conditions, or the adoption or use of one or more practices, means, methods, operations, or processes, reasonably necessary or appropriate to provide safe or healthful employment and places of employment." 29 U.S.C. § 652(8) (emphasis added). The Supreme Court has interpreted this provision to require that, before the promulgation of any permanent health standard, OSHA make a threshold finding that a significant risk of material health impairment exists at the current levels of exposure to the toxic substance in question, Benzene, 448 U.S. at 614-15, 642, 100 S.Ct. at 2850-51, 2864; ATMI, 452 U.S. at 505-06, 101 S.Ct. at 2488-89, "and that a new, lower standard is therefore `reasonably necessary or appropriate to provide safe or healthful employment and places of employment.'" Benzene, 448 U.S. at 615, 100 S.Ct. at 2850. OSHA is not entitled to regulate any risk, only those which present a "significant" risk of "material" health impairment. Id. at 641-42, 100 S.Ct. at 2863-64. OSHA must therefore determine: (1) what health impairments are "material," Texas Independent Ginners, 630 F.2d at 407, and (2) what constitutes a "significant" risk of such impairment, Benzene, 448 U.S. at 641-42, 655, 100 S.Ct. at 2863-64, 2870-71. Moreover, OSHA ultimately bears the burden of proving by substantial evidence that such a risk exists and that the proposed standard is necessary. Id. at 653, 100 S.Ct. at 2869-70. The agency "has no duty to calculate the exact probability of harm," id. at 655, 100 S.Ct. at 2870-71, or "to support its finding that a significant risk exists with anything approaching scientific certainty," id. at 656, 100 S.Ct. at 2871. However, OSHA must provide at least an estimate of the actual risk associated with a particular toxic substance, see Public Citizen Health Research Group v. Tyson, 796 F.2d 1479, 1502-03 (D.C. Cir. 1986), and explain in an understandable way why that risk is significant. Benzene, 448 U.S. at 646, 100 S.Ct. at 2866. In past rulemakings, OSHA has satisfied this requirement by estimating either the number of workers likely to suffer the effects of exposure or the percentage of risk to any particular worker. See ATMI, 452 U.S. at 503, 505 n. 25, 101 S.Ct. at 2487-88, 2870-71 n. 25; Building Constr. Trades Dept., AFL-CIO v. Brock, 838 F.2d 1258, 1263 (D.C. Cir. 1988); Public Citizen, 796 F.2d at 1502; ASARCO, Inc. v. OSHA, 746 F.2d 483, 488 (9th Cir. 1984).

Once OSHA finds that a significant risk of material health impairment exists at current exposure levels for a given toxic substance, any standard promulgated to address that risk must comply with the requirements of section 6(b)(5) of the OSH Act. 29 U.S.C. § 655(b)(5). That section provides that the agency

in promulgating standards dealing with toxic materials or harmful physical agents under this subsection, shall set the standard which most adequately assures, to the extent feasible, on the basis of the best available evidence, that no employee will suffer material impairment of health or functional capacity even if such employee has regular exposure to the hazard dealt with by such standard for the period of his working life. Development of standards under this subsection shall be based upon research, demonstrations, experiments, and such other information as may be appropriate. In addition to the attainment of the highest degree of health and safety protection for the employee, other considerations shall be the latest available scientific data in the field, the feasibility of the standards, and experience gained under this and other health and safety laws. Whenever practicable, the standard promulgated shall be expressed in terms of objective criteria and of the performance desired.

Id. (emphasis added). In other words, section 6(b)(5) mandates that the standard adopted "prevent material impairment of health to the extent feasible." ATMI, 452 U.S. at 512, 101 S.Ct. at 2492 (emphasis omitted).

1. Material Impairment

In this rulemaking, OSHA grouped the 428 substances into eighteen categories by the primary health effects of those substances, for example, neuropathic effects, sensory irritation, and cancer. See 54 Fed. Reg. at 2402-03. Industry petitioners charge that for several categories of substances OSHA failed to adequately justify its determination that the health effects caused by exposure to these substances are "material impairments." We disagree.

Petitioners cite the category of "sensory irritation" as a particularly egregious example. Id. at 2434. At the beginning of the discussion for each category, the agency summarized the types of health effects within that category, and discussed why those effects constituted "material impairments." The "Description of Health Effects" for the "sensory irritation" category includes the following discussion:

The symptoms of sensory irritation include stinging, itching, and burning of the eyes, tearing (or lacrimation), a burning sensation in the nasal passages, rhinitis (nasal inflammation), cough, sputum production, chest pain, wheezing, and dyspnea (breathing difficulty). . . .

These effects may cause severe discomfort and can be seriously disabling, as is the case with dyspnea or wheezing. The tearing and eye irritation associated with exposure to sensory irritants are often severe and can be as disabling as the weeping caused by exposure to tear gas. In addition to these primary effects, workers distracted by material irritant effects are more likely than nonexposed workers to have accidents and thus to endanger both themselves and others. (These adverse health effects also clearly have substantial productivity impacts.)

. . . .

During the rulemaking, the question arose as to the level of irritation that constitutes a significant risk of material impairment. . . . Some commenters were of the opinion that transient irritant effects should not be considered material impairment of health. . . .

Most commentators, however, recommended that these signs and symptoms be regarded as material health impairments . . . .

. . . .

. . . [A]ccording to NIOSH[], sensory irritants interfere with job performance and safety, cause inflammation, may increase the victim's susceptibility to other irritants and infectious agents, lead to permanent injury or dysfunction, or permit greater absorption of hazardous substances.

. . . .

OSHA concludes that exposure limits are needed for those substances for which PELs are being established in this rulemaking to protect against sensory irritant effects that result in objective signs of irritation, such as coughing, wheezing, conjunctivitis, and tearing. Such levels of mucous membrane irritation may require medical treatment, adversely affect the well-being of employees, and place the affected individuals at risk from increased absorption of the substance and decreased resistance to infection. Exposing workers repeatedly to irritants at levels that cause subjective irritant effects[] may cause workers to become inured to the irritant warning properties of these substances and thus increase the risk of overexposure.

54 Fed.Reg. at 2444-45 (citations omitted). In addition, in the more general discussion of OSHA's approach to this rulemaking, OSHA also recognized that

irritation also covers a spectrum of effects, some serious and some trivial. Hence, complaints of minor irritation would not in and of itself constitute material impairment.

In addition, OSHA would weigh irritation with physical manifestations more heavily than irritation with purely subjective responses. This does not mean that purely subjective responses would not constitute material impairment. That judgment would depend on the magnitude of the irritation.

Id. at 2362. We interpret this explanation as indicating that OSHA finds that although minor irritation may not be a material impairment, there is a level at which such irritation becomes so severe that employee health and job performance are seriously threatened, even though those effects may be transitory. We find this explanation adequate. OSHA is not required to state with scientific certainty or precision the exact point at which each type of sensory or physical irritation becomes a material impairment. Moreover, section 6(b)(5) of the Act charges OSHA with addressing all forms of "material impairment of health or functional capacity," and not exclusively "death or serious physical harm" or "grave danger" from exposure to toxic substances. See 29 U.S.C. § 654(a)(1), 655(c). Overall, we find that OSHA's determinations of what constitute "material impairments" are adequately explained and supported in the record.

2. Significant Risk

However, the agency's determination of the extent of the risk posed by individual substances is more problematic. "No one could reasonably expect OSHA to adopt some precise estimate of fatalities likely from a given exposure level, and indeed the Supreme Court has said that the agency has `no duty to calculate the exact probability of harm.'" International Union, UAW v. Pendergrass, 878 F.2d 389, 392 (D.C. Cir. 1989) (quoting Benzene, 448 U.S. at 655, 100 S.Ct. at 2870-71). Nevertheless, OSHA has a responsibility to quantify or explain, at least to some reasonable degree, the risk posed by each toxic substance regulated. See id. ("OSHA necessarily seeks to quantify the risk posed by each toxic threat." (Emphasis added.)); see also Benzene, 448 U.S. at 614-15, 100 S.Ct. at 2849-50 ("We agree . . . that § 3(8) requires the Secretary to find, as a threshold matter, that the toxic substance in question poses a significant health risk. . . ." (Emphasis added.)). Otherwise, OSHA has not demonstrated, and this court cannot evaluate, how serious the risk is for any particular substance, or whether any workers will in fact benefit from the new standard for any particular substance. If each of these 428 toxic substances had been addressed in separate rulemakings, OSHA would clearly have been required to estimate in some fashion the risk of harm for each substance. OSHA is not entitled to take short-cuts with statutory requirements simply because it chose to combine multiple substances in a single rulemaking.

However, OSHA's discussions of individual substances generally contain no quantification or explanation of the risk from that individual substance. The discussions of individual substances contain summaries of various studies of that substance and the health effects found at various levels of exposure to that substance. However, OSHA made no attempt to estimate the risk of contracting those health effects. Instead, OSHA merely provided a conclusory statement that the new PEL will reduce the "significant" risk of material health effects shown to be caused by that substance, see, e.g., 54 Fed.Reg. at 2508 (bismuth telluride), without any explanation of how the agency determined that the risk was significant. However, OSHA did make a generic finding that the Air Contaminants Standard as a whole would prevent 55,000 occupational illnesses and 683 deaths annually. Id. at 2725.

Moreover, a determination that the new standard is "reasonably necessary or appropriate," 29 U.S.C. § 652(8), and that it is the standard that "most adequately assures . . . that no employee will suffer material impairment of health or functional capacity," id. § 655(b)(5), necessarily requires some assessment of the level at which significant risk of harm is eliminated or substantially reduced. See Benzene, 448 U.S. at 653, 100 S.Ct. at 2869-70. Yet, with rare exceptions, the individual substance discussions in the Air Contaminants Standard are virtually devoid of reasons for setting those individual standards. In most cases, OSHA cited a few studies and then established a PEL without explaining why the studies mandated the particular PEL chosen. For example, the PEL for bismuth telluride appears to be based on a single study that showed almost no effects of any kind in animals at several times that concentration. 54 Fed.Reg. at 2508. Similarly, the PEL for ferrovanadium dust was based on pulmonary changes at exposure levels many hundreds of times higher than OSHA's new standard. Id. at 2510. See also, e.g., iron pentacarbonyl, id. at 2412; cesium hydroxide, id. at 2455; iron salts, id. at 2466; ethylene dichloride, id. at 2484; sulfur tetrafluoride, id. at 2526. For some substances, OSHA merely repeated a boilerplate finding that the new limit would protect workers from significant risk of some material health impairment. For example, OSHA did not cite any studies whatsoever for its aluminum welding fume standard, id. at 2554, or its vegetable oil mist standard, id. at 2601. See also, e.g., starch, id. at 2599-2600.

"While our deference to the agency is at a peak for its choices among scientific predictions, we must still look for some articulation of reasons for those choices." Pendergrass, 878 F.2d at 392 (emphasis added).

Explicit explanation for the basis of the agency's decision not only facilitates proper judicial review but also provides the opportunity for effective peer review, legislative oversight, and public education. This requirement is in the best interest of everyone, including the decision-makers themselves. If the decision-making process is open and candid, it will inspire more confidence in those who are affected. Further, by opening the process to public scrutiny and criticism, we reduce the risk that important information will be overlooked or ignored.

AFL-CIO v. Marshall, 617 F.2d at 651-52. Mere conclusory statements, such as those made throughout the Air Contaminants Standard, are simply inadequate to support a finding of significant risk of material health impairment.

On the other hand, OSHA established PELs for carbon tetrachloride and vinyl bromide, both carcinogens, at levels where OSHA itself acknowledged that the risk of material health impairment remained significant. 54 Fed.Reg. at 2679-80, 2694. For carbon tetrachloride, OSHA stated that at the new level, "residual risk continues to be significant . . . 3.7 excess deaths per 1,000 workers exposed over their working lifetimes." Id. at 2680. For vinyl bromide, OSHA stated that the new PEL "will not eliminate this significant risk, because . . . residual risk [at the new level] is 40 excess deaths per 1,000 exposed workers . . . [and thus] is clearly significant." Id. at 2694. The only explanation given by OSHA in the final rule for setting its standard where a significant risk of material health impairment remains was that the time and resource constraints of attempting to promulgate an air contaminants standard of this magnitude prevented detailed analysis of these substances. See id. at 2363, 2694. OSHA did not claim in the final rule that the PELs for these two substances were necessary because of feasibility concern.

The agency's response to this criticism is unpersuasive. OSHA first contends that quantitative risk analysis using mathematical models like the ones developed for carcinogens was impossible for this rulemaking because no such models exist for non-carcinogens.

Dose-response models[] have often been used in the quantitative assessment of the risks associated with exposures to carcinogenic substances. However, less scientific effort has been devoted to models to be used with non-carcinogenic substances. Mathematically precise methods to establish the true no-effect level or to define the dose-response curves have not been developed for most of the more than 400 substances involved in this rulemaking.

Most of the scientific work that has been done was designed to identify lowest observed effect or no-effect levels for a variety of acute effects. . . . It is possible to use these data, combined with professional judgment and OSHA's expertise and experience, to determine that significant risk exists at current levels of exposure and that a reduction in these levels will substantially reduce this risk of material impairment of health.

54 Fed.Reg. at 2399-2400. Yet, in several previous rulemakings, OSHA apparently succeeded in determining how many workers were exposed to a particular substance or how much risk would be alleviated by a new standard, even though those particular substances were not carcinogens. See United Steelworkers, 647 F.2d at 1245-51 (lead poisoning); AFL-CIO v. Marshall, 617 F.2d at 646 (byssinosis caused by cotton dust); see also Building Constr., 838 F.2d at 1263 (asbestosis and cancer). It is therefore unclear whether the lack of a method to quantitatively assess the risk for noncarcinogens is a cause or a result of the agency's approach. In this rulemaking, OSHA concluded that current exposure to 428 substances posed a "significant" risk of material health impairment, and that its new standards were required for most of these substances to eliminate or substantially reduce that risk. It is not unreasonable to require that the agency explain how it arrived at that determination, and, indeed, this is precisely what Congress required.

The agency further claims that no quantification was required because OSHA's final standards "`fall within a zone of reasonableness.'" OSHA Brief at 40-41 (quoting United Steelworkers, 647 F.2d at 1207). However, without any quantification or any explanation, this court cannot determine what that "zone of reasonableness" is or if these standards fall within it.

OSHA also responds by noting that it incorporated "uncertainty" or "safety" factors into many PELs. However, OSHA did not use a uniform safety factor, but instead claims to have made a case-by-case assessment of the appropriate safety factor. 54 Fed.Reg. at 2398-99. "Studies are often of small size and, since there is a large variation in human susceptibility, a study because of its small size may not demonstrate an effect that actually exists. . . . For this reason, it is not uncommon to set a limit below that level which the study may have indicated showed no effect." Id. at 2365. OSHA claims that use of such uncertainty factors "has been the standard approach for recommending exposure limits for non-carcinogens by scientists and health experts in the field for many years." Id. In this rulemaking, the difference between the level shown by the evidence and the final PEL is sometimes substantial. We assume, because it is not expressly stated, that for each of those substances OSHA applied a safety factor to arrive at the final standard. Nevertheless, the method by which the "appropriate" safety factor was determined for each of those substances is not explained in the final rule.

We find OSHA's use of safety factors in this rulemaking problematic. First, OSHA's use of safety factors in this rulemaking is very similar to the approach criticized by the Supreme Court in Benzene. Second, even assuming that the use of safety factors is permissible under the Act and Benzene, application of such factors without explaining the method by which they were determined, as was done in this case, is clearly not permitted.

From OSHA's description, safety factors are used to lower the standard below levels at which the available evidence shows no significant risk of material health impairment because of the possibility that the evidence is incorrect or incomplete; i.e., OSHA essentially makes an assumption that the existing evidence does not adequately show the extent of the risk. That may be a correct assumption, but beyond a general statement that the use of safety factors is common in the scientific community, OSHA did not indicate how the existing evidence for individual substances was inadequate to show the extent of the risk from those substances. Such a rationale is very reminiscent of the "benefits are likely to be appreciable" rationale rejected in Benzene as insufficient to satisfy the agency's obligations under the OSH Act. In Benzene, the Supreme Court noted that

[t]he evidence in the administrative record of adverse effects of benzene exposure at 10 ppm [the former PEL] is sketchy at best. OSHA noted that there was "no dispute" that certain nonmalignant blood disorders, evidenced by a reduction in the level of red or white cells or platelets in the blood, could result from exposures of 25-40 ppm. It then stated that several studies had indicated that relatively slight changes in normal blood values could result from exposures below 25 ppm and perhaps below 10 ppm. OSHA did not attempt to make any estimate based on these studies of how significant the risk of nonmalignant disease would be at exposures of 10 ppm or less. Rather, it stated that because of the lack of data concerning the linkage between low-level exposures and blood abnormalities, it was impossible to construct a dose-response curve at this time. OSHA did conclude, however, that the studies demonstrated that the current 10 ppm exposure limit was inadequate to ensure that no single worker would suffer a nonmalignant blood disorder as a result of benzene exposure. Noting that it is "customary" to set a permissible exposure limit by applying a safety factor of 10-100 to the lowest level at which adverse effects had been observed, the Agency stated that the evidence supported the conclusion that the limit should be set at a point "substantially less than 10 ppm" even if benzene's leukemic effects were not considered. OSHA did not state, however, that the nonmalignant effects of benzene exposure justified a reduction in the permissible exposure limit to 1 ppm.

Benzene, 448 U.S. at 631-32, 100 S.Ct. at 2858-59 (footnotes and citation omitted; emphasis added). Comparing OSHA's rationale for using safety factors in this rulemaking with the Court's discussion of their use in the Benzene case, we find little appreciable difference.

The Supreme Court in Benzene did recognize that absolute scientific certainty may be impossible when regulating on the edge of scientific knowledge, and that "so long as they are supported by a body of reputable scientific thought, the Agency is free to use conservative assumptions in interpreting the data . . ., risking error on the side of overprotection rather than underprotection." Id. at 656, 100 S.Ct. at 2871. However, the Court also discussed the use of monitoring and medical testing as a "backstop," permitting the agency to "keep a constant check on the validity of the assumptions made in developing the permissible exposure limit, giving it a sound evidentiary basis for decreasing the limit if it was initially set too high." Id. at 658, 100 S.Ct. at 2872; see also National Cottonseed Prods. Ass'n v. Brock, 825 F.2d 482, 486 (D.C. Cir. 1987), cert. denied, 485 U.S. 1020, 108 S.Ct. 1573, 99 L.Ed.2d 889 (1988).

The lesson of Benzene is clearly that OSHA may use assumptions, but only to the extent that those assumptions have some basis in reputable scientific evidence. If the agency is concerned that the standard should be more stringent than even a conservative interpretation of the existing evidence supports, monitoring and medical testing may be done to accumulate the additional evidence needed to support that more protective limit. Benzene does not provide support for setting standards below the level substantiated by the evidence. Nor may OSHA base a finding of significant risk at lower levels of exposure on unsupported assumptions using evidence of health impairments at significantly higher levels of exposure. Benzene, 448 U.S. at 656-58, 100 S.Ct. 2871-72; Texas Indep. Ginners, 630 F.2d at 409. Overall, OSHA's use of safety factors in this rulemaking was not adequately explained by this rulemaking record.

More generally, OSHA defends its failure to make more specific findings for each individual substance, as well as its decision to set the standards for several substances at levels where significant risks of material health impairment remain, by citing its authority to set priorities, 29 U.S.C. § 655(g), and the discretion permitted the agency in making policy decisions. There were warning signs of this position in the Notice of Proposed Rulemaking, where OSHA expressed its view that "to review and regulate many substances in a reasonable period requires some narrowing of the issues, focus of analysis, and reducing the length of the discussions in the preamble." 53 Fed.Reg. at 20963. Indeed, the agency stated that "[t]he success of a project to regulate a large backlog of chemicals for which there is a generally recognized need for new or improved employee protection requires some recognition of the need for agency flexibility in several areas," for example, "in less detailed discussion for each substance." Id. Moreover, the agency stated that

[i]n response to both the court challenges and the need to face difficult issues, OSHA has engaged in detailed and extensive analyses. These have resulted in lengthier preamble discussions and in-depth analyses for all issues. . . .

Now that OSHA has reviewed these issues in depth several times, has experience "gained under this * * * law" (sec. 6(b)(5)) on these issues, and has had its analysis upheld in the Courts, somewhat less detailed chemical-by-chemical analyses should be appropriate. The accumulated judicial guidance and agency experience reduces the need for as extensive a discussion of some of the issues.

Id. at 20964. This implies that OSHA need no longer perform detailed analysis and explanation when promulgating PELs because the agency's analysis for other substances has been upheld in prior rulemakings. Besides displaying more than a touch of hubris, this passage reveals a fundamental misperception of the OSH Act and the caselaw interpreting that act.

While OSHA has probably established that most or all of the substances involved do pose a significant risk at some level, it has failed to establish that existing exposure levels in the workplace present a significant risk of material health impairment or that the new standards eliminate or substantially lessen the risk.

C. FEASIBILITY

The Supreme Court has defined "feasibility" as "`capable of being done, executed, or effected,'" ATMI, 452 U.S. at 508-09, 101 S.Ct. at 2490-91 (quoting Webster's Third New International Dictionary 831 (1976)), both technologically and economically, National Cottonseed, 825 F.2d at 487; United Steelworkers, 647 F.2d at 1264; see also ATMI, 452 U.S. at 513 n. 31, 101 S.Ct. at 2492 n. 31 ("[A]ny standard that was not economically or technologically feasible would a fortiori not be `reasonably necessary or appropriate' under the Act." (Emphasis added.)). Again, the burden is on OSHA to show by substantial evidence that the standard is feasible, United Steelworkers, 647 F.2d at 1264-67, although OSHA need not prove feasibility with scientific certainty, id. at 1266. Despite OSHA's repeated claims that it made feasibility determinations on an industry-by-industry basis, it is clear that the agency again proceeded "generically."

1. Technological Feasibility

To show that a standard is technologically feasible, OSHA must demonstrate "that modern technology has at least conceived some industrial strategies or devices which are likely to be capable of meeting the PEL and which the industries are generally capable of adopting." United Steelworkers, 647 F.2d at 1266. Further, "the undisputed principle that feasibility is to be tested industry-by-industry demands that OSHA examine the technological feasibility of each industry individually." Id. at 1301. Courts have remanded OSHA determinations where the agency has not sufficiently analyzed the abilities of different industries to meet proposed standards. See Building Constr., 838 F.2d at 1272-73; Industrial Union Dept., AFL-CIO v. Hodgson, 499 F.2d 467, 480-81 (D.C. Cir. 1974).

In this rulemaking, OSHA first identified the primary air contaminant control methods: Engineering controls are methods such as ventilation, isolation, and substitution. 54 Fed.Reg. at 2789. Complementing the engineering controls are work practices and administrative reforms (e.g., housekeeping, material handling or transfer procedures, leak detection programs, training, and personal hygiene). Id. at 2789-90. Finally, personal protective equipment such as respirators and gloves may become necessary when these other controls are not fully effective. Id. at 2790.

OSHA then organized its discussion of technological feasibility by industry sector using the Standard Industrial Classification (SIC) groupings. The SIC codes classify by type of activity for purposes of promoting uniformity and comparability in the presentation of data. As the codes go from two and three digits to four digits, the groupings become progressively more specific. For example, SIC Code 28 represents "Chemicals and Allied Products," SIC Code 281 represents "Industrial Inorganic Chemicals," and SIC Code 2812 includes only "Alkalies and Chlorine." OSHA primarily relied on the more general two-digit codes in its feasibility analysis. For most of the SIC codes discussed, OSHA provided only a general description of how generic engineering controls might be used in a given sector. Then, relying on this generic analysis, OSHA concluded

that existing engineering controls are available to reduce exposure levels to the new levels.

In reviewing the comments and hearing testimony on the technological feasibility of achieving the PELs and other limits, OSHA has found that for the overwhelming majority of situations where air contaminants are encountered by workers, compliance can be achieved by applying known engineering control methods and work practice improvements.

54 Fed.Reg. at 2789. However, OSHA made no attempt to show the ability of technology to meet specific exposure standards in specific industries. Except for an occasional specific conclusion as to whether a particular process control could meet a particular PEL, OSHA merely presented general conclusions as to the availability of these controls in a particular industry. See, e.g., id. at 2802-03 (SIC 32 — Stone, Clay, and Glass Products).

OSHA correctly notes that all it need demonstrate is "a general presumption of feasibility for an industry." United Steelworkers, 647 F.2d at 1266 (second emphasis added); see also ASARCO, 746 F.2d at 496. However, as this quote indicates, "a general presumption of feasibility" refers to a specific industry-by-industry determination that a "typical firm will be able to develop and install engineering and work practice controls that can meet the PEL in most of its operations." United Steelworkers, 647 F.2d at 1272. OSHA can prove this "by pointing to technology that is either already in use or has been conceived and is reasonably capable of experimental refinement and distribution within the standard's deadlines." Id. Only when OSHA has provided such proof for a given industry does there arise "a presumption that industry can meet the PEL without relying on respirators, a presumption which firms will have to overcome to obtain relief in any secondary inquiry into feasibility." Id.

For example, in United Steelworkers, the court analyzed whether the agency had provided enough information about the relative abilities of different industries to meet the standard. The court upheld OSHA's rationale with regard to lead smelting industries because OSHA made industry-specific findings and identified specific technologies capable of meeting the proposed limit in industry specific process operations. Id. at 1278-89. As to numerous other industries, however, the court remanded because OSHA's findings lacked a detailed industry- or operation-specific analysis. For example, in discussing one such industry, nonferrous foundries, the court unfavorably compared OSHA's analysis of that industry to that of the lead-smelting industry, stating:

[OSHA] relies on the testimony of Gary Mosher of the American Foundrymen's Society . . . but concedes that Mosher offered no opinion whether such exhaust systems could meet even the [higher] standard. It lists conventional devices . . . which, it asserts, can be refined enough over time to meet the PEL. But it cites no record evidence at all to support this view. It analyzes in modest detail such specific operations as molding, melting, and pouring, and cites [a] report for suggestions on technology that will be necessary to reduce lead exposure, but cites no evidence whatever on the ability of such technology to meet any specific PEL in these operations. . . . Thus OSHA has hardly asserted, much less offered substantial evidence for, the feasibility of the PEL for these foundries.

647 F.2d at 1293-94 (citations footnotes omitted; emphasis added). Thus, it is clear that the concept of "a general presumption of feasibility" does not grant OSHA a license to make overbroad generalities as to feasibility or to group large categories of industries together without some explanation of why findings for the group adequately represent the different industries in that group. We find that OSHA has not established the technological feasibility of the 428 PELs in its revised Air Contaminants Standard.

2. Economic Feasibility

Nor has OSHA adequately demonstrated that the standard is economically feasible. OSHA must "provide a reasonable assessment of the likely range of costs of its standard, and the likely effects of those costs on the industry," id. at 1266, so as to "demonstrate a reasonable likelihood that these costs will not threaten the existence or competitive structure of an industry, even if it does portend disaster for some marginal firms," id. at 1272. The determination of economic feasibility is governed by the same principles as technological feasibility. It must be supported by substantial evidence and OSHA must demonstrate its applicability to the affected industries. See id. at 1301 n. 160.

In this rulemaking, although OSHA ostensibly recognized its responsibility "to demonstrate economic feasibility for an industry," 54 Fed.Reg. at 2367 (emphasis added), the agency nevertheless determined feasibility for each industry "sector" (i.e., two-digit SIC code), without explaining why such a broad grouping was appropriate. Id. OSHA's economic feasibility determinations therefore suffer from the same faults as its technological feasibility findings. Indeed, it would seem particularly important not to aggregate disparate industries when making a showing of economic feasibility. OSHA admits that its economic feasibility conclusions only "have a high degree of validity on a sector basis," id., as opposed to a subsector or more industry-specific basis. See also id. at 2374 ("OSHA concludes that this approach is accurate on an industry sector by industry sector basis for individual processes."). OSHA then stated that "[t]he costs are sufficiently low per sector to demonstrate feasibility not only for each sector but also for each subsector." Id. at 2367.

However, reliance on such tools as average estimates of cost can be extremely misleading in assessing the impact of particular standards on individual industries. Analyzing the economic impact for an entire sector could conceal particular industries laboring under special disabilities and likely to fail as a result of enforcement. Moreover, for some substances, OSHA failed even to analyze all the affected industry sectors. See, e.g., id. at 2686-89 (perchloroethylene). We find that OSHA has not met its burden of establishing that its 428 new PELs are either economically or technologically feasible.

D. THE PERC EXAMPLE

OSHA's analysis of perchloroethylene (perc) is a prime example of the problems with OSHA's approach to this rulemaking. Perc is a widely used solvent in the drycleaning and industrial degreasing industries. Id. at 2686. Petitioners, the International Fabricare Institute ("IFI") and the Halogenated Solvents Industry Alliance, argue that OSHA's determination that perc presents a significant cancer risk is not supported by substantial evidence. On the other hand, the union argues that the final PEL adopted by OSHA, 25 ppm, though lower than the 50 ppm limit originally proposed, still leaves workers exposed to a significant cancer risk. Indeed, as OSHA recognized, the quantitative risk assessment on which the agency relied showed that even at an exposure level of 10 ppm, an excess risk of 6.4 deaths per 1000 workers would remain, id. at 2687, which by OSHA's own admission is considered significant. OSHA, however, cited feasibility concerns for its decision not to lower the PEL even further. Id. at 2688. If supported by substantial evidence in the record, this would satisfy the requirements of the OSH Act.

Although OSHA stated that it "does not believe that information in the record at the present time demonstrates that it is feasible to reduce exposures to lower levels," id., OSHA's feasibility analysis for perc is grossly inadequate. For technological feasibility, OSHA limited its discussion to showing that its new PEL of 25 ppm was achievable. OSHA stated that 25 ppm can be achieved with newer equipment and engineering and work practice controls. Id. OSHA also stated that "a significant percentage of operations, including smaller operations, have installed" the newer equipment, and that the industry as a whole "is gradually replacing older equipment with newer equipment." Id.; see also id. at 2811-12. However, there is no explanation or evidence cited in the final rule to support the proposition that an even lower PEL is not technologically feasible.

On the other hand, OSHA's economic feasibility determination for perc cannot support either the new PEL of 25 ppm or the agency's decision not to set an even lower PEL. OSHA used the two-digit SIC code, SIC 72 — Personal Services, to define the industries affected by the perc standard. This creates two problems. First, drycleaning is the only industry in SIC 72 affected by the perc standard. SIC 72 covers numerous other industries, including funeral services, shoe repairs, barber and beauty shops, and photography studios. See id. at 2763. Nevertheless, OSHA took the costs of compliance with the new perc standard, which would be borne only by the drycleaning industry subsector (SIC code 7216), and compared those costs to the profits and sales of the entire personal services sector (SIC 72). As a result, OSHA must have significantly understated the costs of compliance for the drycleaning industry. Indeed, petitioners claim that the actual economic impact on this industry would be more than ten times OSHA's estimate.

Moreover, while the drycleaning industry received at least some feasibility analysis for perc, the other major user of that chemical, industrial degreasing operations, received none. This industry is not in SIC 72, which was the only industry sector reviewed for technological or economic feasibility for the new perc standard. Therefore, OSHA clearly has not fulfilled its duty to examine the feasibility of its perc standard for each affected industry.

Accordingly, while OSHA has determined that significant risk remains for exposure to perc at a level of 25 ppm, it is not clear from the record that 25 ppm is the lowest feasible level. The union claims that it would be feasible to set the standard lower than 25 ppm, perhaps as low as 10 ppm. Industry, on the other hand, claims that even 25 ppm is not feasible. From the record and final rule as presented, it is impossible for this court to determine which claim is correct.

E. OTHER UNION ISSUES

Union petitioners raise several other issues that deserve comment.

1. Use of ACGIH Recommendations

The union challenges OSHA's use of the ACGIH recommendations on two grounds. First, OSHA only considered for this rulemaking those substances for which the ACGIH has recommended a limit more protective than the existing PEL or for which there was no existing PEL. 54 Fed. Reg. at 2372; 53 Fed.Reg. at 20967. The union argues that this decision to limit the scope of this rulemaking in such a way was inconsistent with the agency's duty to set the standards "which most adequately assure . . . that no employee will suffer material impairment of health or functional capacity" from exposure to toxic substances. See 29 U.S.C. § 655(b)(5). We disagree.

As previously noted, we find nothing in the OSH Act that prohibits OSHA from combining multiple substances in one rulemaking, as long as the statutory requirements are met for each substance. Neither do we find a requirement that OSHA include all possible substances in one rulemaking. OSHA has never claimed that this Air Contaminants Standard constituted the total universe of substances needing regulation, and it seems reasonable that some limit needed to be set as to what substances could be considered in this rulemaking. The list of ACGIH recommendations is a rational choice as the source for that limitation. The ACGIH recommendations are clearly well known to industry and the safety and health community. Therefore, we find that the agency's choice to so limit this rulemaking is a valid exercise of OSHA's authority to set priorities for rulemaking. See id. § 655(g). The Act is sufficiently flexible to allow OSHA "to initially determine whether or not there will be a standard . . . [and] to process higher priority standards more quickly." National Congress of Hispanic Am. Citizens v. Usery, 554 F.2d 1196, 1199 (D.C. Cir. 1977).

Petitioners also challenge OSHA's extensive use of the ACGIH recommendations for individual substances. Petitioners claim that OSHA did no more than adopt wholesale the ACGIH recommendations without independently analyzing the evidence supporting those recommendations. It is not clear from the record if OSHA independently reviewed the individual studies cited by the ACGIH, or merely the ACGIH summary of that data. See 53 Fed.Reg. at 20965 ("OSHA has concentrated its efforts on reviewing summaries of the major studies, with emphasis on the literature used to support the exposure limits proposed by NIOSH and ACGIH. . . ."). However, we do not believe that OSHA is required to independently research all aspects of its rules. The Act authorizes the agency to employ expert consultants, 29 U.S.C. § 656(c), and OSHA is entitled to rely on such consultants, as well as studies in the scientific literature, as a basis for its PELs. See United Steelworkers, 647 F.2d at 1217. To that extent, OSHA is entitled to rely on the recommendations and documentation of the ACGIH, as it may rely on any other consulting organizations.

Use of such consultants, however, does not relieve OSHA of the responsibility for making detailed findings, with adequate explanations, for all statutory criteria. The recommendations of a consultant are not always based on the criteria required by the statute. As it is ultimately OSHA's responsibility to make those statutory findings, OSHA must determine, based on the "best available evidence," 29 U.S.C. § 655(b)(5), if those statutory criteria have been met, and then set forth the analysis behind that determination in an understandable way.

In this rulemaking, petitioners state that the ACGIH recommendations are not based on the statutory criteria and that OSHA did nothing more than adopt those recommendations without any independent analysis of whether the evidence supporting those recommendations satisfied the statutory criteria. We express no opinion on this claim. However, the dearth of explanation in the rulemaking record for these 428 PELs makes it difficult to determine how the agency arrived at its conclusions. We do note that not all final PELs conformed to the ACGIH recommendation, though approximately ninety percent were identical. On remand, OSHA will be required to carefully review the evidence supporting each of the 428 PELs, and this concern should be addressed at that time.

2. Exclusion of Monitoring and Medical Surveillance

The union also challenges OSHA's decision to defer issuing standards for monitoring and medical surveillance of the new PELs until a later rulemaking. Section 6(b)(7) of the OSH Act provides:

Where appropriate, such standard shall also prescribe suitable protective equipment and control or technological procedures to be used in connection with such hazards and shall provide for monitoring or measuring employee exposure at such locations and intervals, and in such manner as may be necessary for the protection of employees. In addition, where appropriate, any such standard shall prescribe the type and frequency of medical examinations or other tests which shall be made available, by the employer at his cost . . . .

29 U.S.C. § 655(b)(7). From this, the union infers that toxic substance standards must always include provisions for monitoring. "Section 6(g) clearly permits the Secretary to set priorities for the use of the agency's resources, and to promulgate standards sequentially." Auchter, 763 F.2d at 738. OSHA argues, and we agree, that this is purely a matter of regulatory priority. See 29 U.S.C. § 655(g).

3. Four-year Compliance Period

The union also challenges OSHA's decision to allow four years, until December 31, 1992, for the implementation of engineering and work practice controls to bring industry into compliance with the new standards, while in the interim permitting compliance through the use of respirators. 54 Fed.Reg. at 2921. As a transitional provision, OSHA specified that employers must continue to achieve the 1971 PELs by adhering to the hierarchy of controls in 29 C.F.R. § 1910.1000(e), as they have been required to do since 1971. In adopting this four-year time period, OSHA stated that the agency's "experience is that for substances of normal difficulty, one to two years is sufficient," 54 Fed.Reg. at 2916, but that a four-year period "takes into account that some employers will have to control several substances and also considers those few substances where compliance may take greater efforts for some employers," id. That conclusory analysis falls short of justifying an across-the-board four-year period of delay, but is fully consistent with OSHA's treatment of this standard as a "generic" standard, without adequate consideration of individual substances or the effect of the new standards on individual industries.

This "generic" four-year compliance period is simply not adequately supported in the record. Unlike other standards where OSHA has exercised its "technology-forcing" authority and required that industries develop the technology to achieve the new standards, see United Steelworkers, 647 F.2d at 1264-65, in this standard, "OSHA's feasibility analysis was based on what industry is already achieving or what could be achieved with standard `off-the-shelf' technology, [and] there are few if any cases where OSHA is attempting to force technology." 54 Fed.Reg. at 2366. If the technology exists and is in many cases already being used, it is difficult to understand why four years is required for the implementation of this standard for all industries. If OSHA's concern was primarily economic feasibility, that too needed to be addressed for each industry or for each appropriate industrial grouping.

To the extent that there may be any unusual situations in which a feasibility problem exists, the OSH Act and the standard itself provide appropriate means of dealing with such problems without resorting to the extreme expedient of an across-the-board four-year compliance period. First, section 6(b)(6) allows an employer to obtain a temporary variance if that employer "is unable to comply with a standard by its effective date because of unavailability of professional or technical personnel or of materials and equipment needed to come into compliance with the standard or because necessary construction or alteration of facilities cannot be completed by the effective date." 29 U.S.C. § 655(b)(6)(A)(i). Furthermore, if the rulemaking record establishes that specific industries will need an extended period of time to comply with PELs for certain substances, OSHA could so provide, just as it allows respirator use after December 1992 for four substances in specified operations. 54 Fed.Reg. at 2335, 2916 (carbon monoxide, carbon disulfide, styrene, and sulfur dioxide). We find insufficient explanation in the record to support this across-the-board four-year delay in implementation of this rule.

IV. CONCLUSION

It is clear that the analytical approach used by OSHA in promulgating its revised Air Contaminants Standard is so flawed that it cannot stand. OSHA not only mislabeled this a "generic" rulemaking, but it inappropriately treated it as such. The result of this approach is a set of 428 inadequately supported standards. OSHA has lumped together substances and affected industries and provided such inadequate explanation that it is virtually impossible for a reviewing court to determine if sufficient evidence supports the agency's conclusions. The individual substances discussed in this opinion are merely examples of what is endemic in the Air Contaminants Standard as a whole.

OSHA does have the authority to set priorities for establishing standards under the OSH Act. See 29 U.S.C. § 655(g). That priority-setting authority permits OSHA to combine rulemaking for multiple substances in one rulemaking, to limit the scope of this rulemaking in a rational manner, and to defer issuance of regulations for monitoring and medical surveillance until a later rulemaking. It does not, however, give OSHA blanket authority to pick and choose what statutory requirements it will follow. The OSH Act mandates that OSHA promulgate the standards that "most adequately" assure that workers will not be exposed to significant risks of material health impairment "to the extent feasible" for the affected industries. Further, section 6(e) and caselaw require OSHA to adequately explain its determinations. Section 6(b) of the Act does not provide an exception to these requirements for administrative convenience. The only exceptions to the strict statutory criteria are the start-up provisions of section 6(a), the applicability of which has long since passed, and the emergency provisions of section 6(c), neither of which are implicated in this case.

Therefore, although we find that the record adequately explains and supports OSHA's determination that the health effects of exposure to these 428 substances are material impairments, we hold that OSHA has not sufficiently explained or supported its threshold determination that exposure to these substances at previous levels posed a significant risk of these material health impairments or that the new standard eliminates or reduces that risk to the extent feasible. OSHA's overall approach to this rulemaking is so flawed that we must vacate the whole revised Air Contaminants Standard.

We have no doubt that the agency acted with the best of intentions. It may well be, as OSHA claims, that this was the only practical way of accomplishing a much needed revision of the existing standards and of making major strides towards improving worker health and safety. Given OSHA's history of slow progress in issuing standards, we can easily believe OSHA's claim that going through detailed analysis for each of the 428 different substances regulated was not possible given the time constraints set by the agency for this rulemaking. Unfortunately, OSHA's approach to this rulemaking is not consistent with the requirements of the OSH Act. Before OSHA uses such an approach, it must get authorization from Congress by way of amendment to the OSH Act. Legislative decisions on the federal level are to be made in the chambers of Congress. It is not for this court to undertake the substantial rewriting of the Act necessary to uphold OSHA's approach to this rulemaking.

Therefore, for the reasons stated above, we VACATE the revised Air Contaminants Standard, and REMAND to the agency.