Justice Devine delivered the opinion of the Court.

Greg W. Abbott, Warren Kenneth Paxton, Attorney Generals of Texas, Daniel T. Hodge, Charles E. Roy, First Assistant Attorney Generals, David C. Mattax, Director of Defense Litigation, Susan Desmarais Bonnen, Walter C. Brocato, James Edward Davis, Randall K. Hill, Cavitt Wendlandt, Office of the Attorney General, Austin TX, for Respondent.

Alan B. Daughtry, Attorney-at-Law, James Juranek, Juranek Law Firm, PLLC, Edward A. Mattingly, Mattingly Law Firm, Houston TX, for Petitioner.

Justice Devine delivered the opinion of the Court.

This condemnation case requires us to review the trial court's exclusion of evidence concerning the government's role in delaying the condemned property's environmental cleanup prior to taking. At trial, the government presented testimony that it would take eight years of cleanup to render the property marketable and that the condemned property's value should be substantially discounted on that basis. In rebuttal, the condemnee offered testimony from two witnesses that the government's condemnation project delayed the property's cleanup. However, at the government's request, the trial court excluded that testimony. The court of appeals affirmed the exclusion. 468 S.W.3d 94 (Tex.App.–Houston [14th Dist.] 2014) (mem.op.). We hold that the trial court's exclusion was an abuse of discretion, and further hold that the exclusion was harmful because it allowed the government to use an eight-year holding period to reduce the property's value without allowing the jury to consider the role the government played in creating that holding period. Accordingly, we reverse the court of appeals' judgment and remand the case for a new trial.

I. Factual Background and Procedural History

A. The Property

Caffe Ribs, Inc. (Caffe) purchased the condemned property from Paul Revere Variable Annuity Insurance Company (Revere) in 1995. At that time, Revere was evaluating the environmental condition of the property. Soil and groundwater sampling conducted at the request of Revere and a previous owner, Weatherford U.S., Inc. (Weatherford), indicated that the property was contaminated. The source and the extent of that contamination, however, were unknown. Given these facts, Caffe and Revere agreed that Revere would continue its evaluation of the property's environmental condition after closing and would take such action as it deemed necessary to remediate the contamination.

In 1996, Revere and Weatherford agreed to cooperate in evaluating and remediating the property's contamination. By February 2000, Revere and Weatherford had identified the source of the contamination and placed the property into the Texas Commission on Environmental Quality's (TCEQ's) Voluntary Cleanup Program.

After the condemned property entered the Voluntary Cleanup Program, Revere and Weatherford began remediation, removing more than 460 cubic yards of contaminated soil in 2001. The TCEQ, however, wasn't satisfied that the extent of the contamination had been fully delineated and requested that Revere and Weatherford do so. In response to the TCEQ's request, Revere and Weatherford had additional soil and groundwater sampling performed.

In January 2003, the State notified Caffe Ribs that it intended to condemn the property, or at least part of it, in connection with the Texas Department of Transportation's project to expand Interstate 10. Initially, the State's revelation did not affect Revere and Weatherford's efforts to delineate and remediate the contamination. In the months following the revelation, additional groundwater sampling was conducted, and additional contaminated soil was removed. By December 2003, Revere and Weatherford believed that the contamination had been fully delineated, and filed an Affected Property Assessment Report (APAR) with the TCEQ outlining the extent of the contamination. In July 2004, the TCEQ responded to Revere and Weatherford's APAR with a request that at least four additional groundwater monitoring wells be installed to more fully delineate the contamination.

Revere and Weatherford, however, were not able to comply with the TCEQ's request due at least in part to the State's impending condemnation. As part of the State's conversion of Caffe's property into a stormwater detention pond, the State requested all existing groundwater monitoring wells on the property be plugged and abandoned, and any new wells be installed after construction was complete. In late 2004, Revere and Weatherford filed an amended APAR that reflected these facts. They also filed a Response Action Plan (“RAP”) containing their proposals for fully remediating the property's contamination. The TCEQ declined to approve either, however, based on its prior unsatisfied request for at least four additional groundwater monitoring wells. The TCEQ also noted that the RAP could not be approved because the existing network of groundwater monitoring wells was to be removed as part of “site construction activities.” According to the TCEQ, Revere and Weatherford's proposed response—a “plume management zone”—required that the zone's network of monitoring wells be in place before TCEQ approval.

B. Caffe I

In May 2005, the State initiated statutory condemnation proceedings against Caffe. Special commissioners were then appointed to value the property. Caffe and the State both objected to the special commissioners' valuation, and the case proceeded to trial. At trial, the State argued that the property's market value was significantly affected by potential environmental liability and remediation costs, but successfully excluded Caffe's proffered evidence that Revere and Weatherford were responsible for such liability and costs. The court of appeals concluded that the exclusion was harmful and remanded the case for a new trial. Caffe I, 328 S.W.3d at 931–33. This appeal concerns the second trial.

C. Caffe II

Before retrial, the State designated Ashby McMullan as its environmental expert to testify that, at the time of taking, the nature and extent of the contamination were uncertain and that it would take eight years to delineate the contamination, complete cleanup, and obtain regulatory approval or closure, as evidenced by a conditional certificate of completion. In conjunction with Mr. McMullan's testimony, the State designated David Dominy as its appraisal expert to testify that, given the property's environmental condition, it would not be marketable for eight years, and accordingly its value should be discounted at an 18% rate over that eight-year holding period. According to Dominy, the property, when clean, would be worth $3,154,150, but the present value, discounted at 18% over eight years, was $803,431.

For its part, Caffe designated Richard Bost, its environmental expert, and Vince Rorick, the former TCEQ project manager responsible for overseeing the property's cleanup, to testify that, beginning in 2004, the State's revelation of the proposed future use of the property as a stormwater detention pond caused delays in progressing toward regulatory closure, and that, as of 2004, the property could have received regulatory closure in as little as one year, but for the change in proposed future use. Caffe designated Rudy Robinson as its appraisal expert to testify that the property's environmental condition did not warrant an adjustment to its market value of $9.9 million. Caffe also designated David Klein, a commercial real estate developer, to testify that the property could be developed in its current condition.

The State filed a pre-trial motion to exclude all testimony concerning its role in delaying the property's remediation on the grounds that such testimony amounted to a claim for “announcement damages” barred by Westgate Ltd. v. State, 843 S.W.2d 448 (Tex.1992), and that such evidence was unreliable. The State also filed a pre-trial motion in limine on the same subject. At the pre-trial conference, the trial court granted the State's motion, concluding that exclusion was required by the “project influence” rule.

At trial, McMullan testified that Revere and Weatherford's APARs and RAP were not approved by the TCEQ because Revere and Weatherford had not delineated the extent of the contamination. He concluded that the most probable timeline for obtaining regulatory closure was eight years, but conceded that it could be shorter if Revere and Weatherford sought to pursue aggressive response actions. According to McMullan, the four steps that remained to be completed to obtain regulatory closure were: (1) TCEQ approval of an APAR; (2) TCEQ approval of a RAP; (3) implementation of the RAP; and (4) completion of a Response Action Completion Report. Dominy testified that the property's value should be significantly discounted over an eight-year holding period. He justified the extended holding period with McMullan's testimony, as well as Revere and Weatherford's lack of “real financial motivation or incentive to go start spending money to do a clean-up.”

To bolster the testimony of their witnesses, the State elicited testimony from Bost, Rorick, and Klein that Revere and Weathereford's APARs and RAP were not approved by the TCEQ, because Revere and Weatherford had not installed the four additional groundwater monitoring wells the TCEQ had requested in July 2004 to more fully delineate the extent of the contamination. However, the State objected when Rorick attempted to explain that “they weren't able to do that because of the freeway.” At the bench conference, Caffe took the position that the State's questioning opened the door for evidence concerning the State's role in delaying the property's remediation. The trial court ruled that “[y]ou can't open the door on project influence,” and instructed Rorick “not to go there.”

Caffe made offers of proof of the testimony it would have elicited from Bost and Rorick. See Tex. R. Evid. 103(a)(2). At the trial court's request, they were made as brief summaries. Id. at 103(c). Bost's proposed testimony was summarized as follows:

Mr. Bost would say, if he were allowed to testify, that at around 2004 when the State told the TCEQ that this property was going to be used as a detention pond, that changed [ ] the comments and questions and requirements for the TCEQ's remediation of this property, [ ] because it was removing the existing well network and requiring placement of other wells. The TCEQ had said, hey, you can't do this until after the State's detention pond project is done. Therefore, when the State is saying, hey, you didn't get it done by the date of taking, Mr. Bost would say there was delay caused by the project.

Rorick's proposed testimony was summarized similarly. In connection with the Rorick offer of proof, Caffe provided the court with a copy of a Rorick affidavit that it had attached to a pre-trial motion. The affidavit stated, in relevant part, that “[t]he State raised the issue of the detention pond construction plan with TCEQ staff prior to 2005,” and that revelation “did affect timing of the administrative process and the pending [Voluntary Cleanup Program] agreement, causing delay.” The affidavit also stated that, as of 2004, the property could have received a conditional certificate of property within one year, if remediation was “aggressively pursued.”

The jury determined the value of the property to be $4,914,480. Caffe timely appealed the jury's verdict, asserting that the trial court committed reversible error by (among other things) excluding evidence concerning the State's role in delaying the property's remediation. The court of appeals affirmed, holding that “[e]ven if the trial court abused its discretion in excluding Bost's and Rorick's proffered testimony, the exclusion was harmless.” 468 S.W.3d at 111.

II. Analysis

We review a trial court's exclusion of evidence under the abuse of discretion standard. Interstate Northborough P'ship v. State, 66 S.W.3d 213, 220 (Tex.2001). A trial court abuses its discretion when it acts without regard for any guiding rules. Downer v. Aquamarine Operators, Inc., 701 S.W.2d 238, 241–42 (Tex.1985). In this case, the trial court excluded the complained–of evidence under the guise of the project-influence rule. Accordingly, that rule is the starting point of our analysis.

A. The Project–Influence Rule

The Texas Constitution provides that “[n]o person's property shall be taken, damaged or destroyed for or applied to public use without adequate compensation being made.” Tex. Const. art. I, § 17. We have effectuated this constitutional imperative by requiring payment of the “market value” of the condemned property—that is, “the price which the property would bring when it is offered for sale by one who desires, but is not obligated to sell, and is bought by one who is under no necessity of buying.” City of Austin v. Cannizzo, 153 Tex. 324, 267 S.W.2d 808, 815 (1954). An impending condemnation project, however, can distort the value of property. See City of Fort Worth v. Corbin, 504 S.W.2d 828, 830–31 (Tex.1974) ; Barshop v. City of Houston, 442 S.W.2d 682, 685 (Tex.1969). The inflationary effects of such a project are referred to as “project enhancement,” Corbin, 504 S.W.2d at 831, while the deflationary effects are referred to as “condemnation blight,” or, perhaps more appropriately, project diminishment. Thurow v. City of Dallas, 499 S.W.2d 347, 348–49 & n. 1 (Tex.Civ.App.–Dallas 1973, writ ref'd n.r.e). Since neither project enhancement nor project diminishment reflects true “market value”—that is, what a willing buyer would pay a willing seller under market conditions—the project-influence rule has evolved to ensure that such components of value are removed from the market-value determination. See, e.g., Corbin, 504 S.W.2d at 830 ; Barshop, 442 S.W.2d at 686. The rule thus provides that any change in property value that results from the government manifesting a definite purpose to take property as part of a governmental project must be excluded from an award of adequate compensation. Corbin, 504 S.W.2d at 831. The rule ensures that the condemnee is made whole, not placed in either a better or worse position than he or she would have enjoyed had there been no condemnation. Id.

While the project-influence rule may be neatly stated, it is not always so neatly applied, as our precedent recognizes. We have previously instructed trial courts that they, not the jury, must make the preliminary determinations that the evidence warrants application of the rule and, if applicable, the date the government manifested a definite purpose to take the condemned property (i.e., the date after which any change in value attributable to the governmental project should be disregarded). Corbin, 504 S.W.2d at 831. We have not, however, instructed trial courts on the exact manner in which the project's influence on the condemned property's value must be corrected, because that depends on the facts of the particular case. While we have long recognized the preferable course to admit evidence, under proper instruction, to permit the jury to eliminate the distorting effect of the project, see, e.g., Morrow v. St. Louis, A. & T.R. Co., 81 Tex. 405, 17 S.W. 44, 44 (1891), we have never foreclosed use of an evidentiary exclusion to eliminate the distorting effect of the project in an appropriate case. We believe the use of a proper instruction, as opposed to an evidentiary exclusion, is particularly appropriate in cases like this, where project diminishment is implicated. In such cases, the value of the condemned property, of course, reflects the deflationary effects of the government's project. If the condemnee is unable to introduce evidence about that deflation, it will be difficult, if not impossible, for the deflation to be factored into the value of the property to arrive at an award of “market value.”

In this case, the trial court enforced the project-influence rule with a sweeping evidentiary exclusion. As just discussed, an evidentiary exclusion is not an essential component of the project-influence rule. Indeed, in this case, the evidentiary exclusion was antithetical to the rule's substantive purpose of ensuring that the compensation award did not reflect the deflationary effects of the government's project. As the United States Supreme Court has observed, “it would be manifestly unjust to permit a public authority to depreciate property values by a threat of the construction of a government project and then to take advantage of this depression in the price which it must pay for the property when eventually condemned.” United States v. Va. Elec. & Power Co., 365 U.S. 624, 636, 81 S.Ct. 784, 5 L.Ed.2d 838 (1961) (citation and internal quotation marks omitted). Yet, the trial court's evidentiary exclusion allowed the State to do just that. It allowed the State to use an eight-year holding period to devalue the property, without allowing the jury to account for the role the State's condemnation project played in that holding period. We therefore conclude that the trial court abused its discretion in applying the project-influence rule. The project-influence rule required the trial court to admit the evidence concerning the State's role in delaying the property's remediation for the very reason the State sought, and the trial court ordered, its exclusion: the evidence was “an attempt to collect damages for a decrease in market value caused by the public announcement of the future detention pond project.”

B. Westgate

As an alternative ground to affirm, the State asserts that the trial court's exclusion was proper under this Court's holding in Westgate, 843 S.W.2d 448. The State's reliance on Westgate is misplaced.

In Westgate, the government commenced statutory condemnation proceedings in September 1988. 843 S.W.2d at 451. The condemnee then counterclaimed for inverse condemnation, seeking to recover lost profits from the four-year period between the announcement of the condemnation project and the government's actual acquisition of the property. Id. In essence, the condemnee asserted that the government's announcement of its project amounted to a de facto taking of his property. In seeking to establish a de facto taking before the de jure taking identified by the statutory condemnation proceeding, the condemnee sought to recover consequential damages in the form of lost profits that otherwise would not have been recoverable. We rejected the condemnee's attempts to do so, recognizing that “publicly targeting a property for condemnation” in the future does not give rise to a de facto taking “unless there is some direct restriction on use of the property.” Id. at 453. We also recognized that the damages sought by the condemnee were “noncompensable consequential damage[s].” Id. at 453 (citing Hubler v. City of Corpus Christi, 564 S.W.2d 816, 821 (Tex.Civ.App.–Corpus Christi 1978, writ ref'd n.r.e.) ).

Caffe neither contends that the announcement of the project amounted to a de facto taking, nor otherwise seeks to recover noncompensable consequential damages, such as lost profits. Caffe instead seeks to recover the hypothetical price that its property would have brought on the date of the State's de jure taking, unaffected by the project. The project-influence rule, not Westgate, therefore, applies. Thurow, 499 S.W.2d at 348 n. 1 (distinguishing project diminishment from a de facto taking).

C. Reliability

The State additionally asserts as an alternative ground for affirming that Bost and Rorick's proffered testimony is unreliable, because it is contradicted by the following “proven facts”: that Revere and Weatherford, not Caffe, had control over the timing of remediation, and that, as of July 2004, Revere and Weatherford had not delineated the contamination, despite knowing about it for fourteen years. We have instructed that “[w]hen expert testimony is involved, courts are to rigorously examine the validity of the facts and assumptions on which the testimony is based.” Whirlpool Corp. v. Camacho, 298 S.W.3d 631, 637 (Tex.2009). When an expert's opinion is predicated on a particular set of facts, those facts need not be undisputed. Hous. Unlimited, Inc. Metal Processing v. Mel Acres Ranch, 443 S.W.3d 820, 833 (Tex.2014). An expert's opinion is only unreliable if it is contrary to actual, undisputed facts. Burroughs Wellcome Co. v. Crye, 907 S.W.2d 497, 499 (Tex.1995).

In this case, Bost and Rorick's proffered testimony is not contrary to actual, undisputed facts. The gravamen of their testimony is that the State's proposed future use of the property as a stormwater detention pond and, specifically, its requirement that the existing network of monitoring wells be removed and only replaced after site construction, delayed Revere and Weatherford's attempts to delineate the contamination after TCEQ requested additional groundwater monitoring wells be installed in July 2004. At most, the State's “proved facts” suggest that Revere and Weatherford may bear some responsibility for the delay in installing those wells. But it does not disprove Bost and Rorick's testimony that the State also bears some responsibility.

D. Harmful Error

The trial court's error in excluding evidence of the State's role in delaying the condemned property's cleanup is reversible only if it probably caused the rendition of an improper judgment. Tex. R. App. P. 44.1(a)(1), 61.1(a)(1). We have recognized the impossibility of establishing a specific test for determining harmful error, and thus have entrusted the matter to the sound discretion of the reviewing court. State v. Cent. Expressway Sign Assocs., 302 S.W.3d 866, 870 (Tex.2009). We have, however, explained that exclusion is likely harmless if the evidence was cumulative, or if the rest of the evidence was so one-sided that the error likely made no difference. Id. In contrast, we have explained that exclusion is likely harmful if evidence is “crucial to a key issue.” Reliance Steel & Aluminum Co. v. Sevcik, 267 S.W.3d 867, 873 (Tex.2008).

The court of appeals concluded that the exclusion of the complained-of evidence was harmless because it did not demonstrate that regulatory closure could have been obtained by the date of taking “but for” the State's interference—an argument Caffe levied at trial. 468 S.W.3d at 110–11. While we may be inclined to agree with the court of appeals on the narrow point that regulatory closure could not have been obtained by the date of taking, the complained-of evidence went far beyond Caffe's argument that regulatory closure could have occurred by the date of taking if Revere and Weatherford pursued aggressive response actions. It went to the validity of the State's eight-year holding period, which was crucial to the only issue before the jury—the property's value. The exclusion was thus harmful, because it allowed the State to use an eight-year holding period to reduce the property's value without allowing the jury to consider the role the State played in creating that holding period.

This harm was compounded by the State's repeated assertions that Revere and Weatherford were simply “trying to spend as little money as possible” or “stalling” in their cleanup of the property. At the parties' first trial, the State successfully excluded all evidence concerning Revere and Weatherford's responsibility for cleaning up the property. The court of appeals recognized that this exclusion “created a false impression that a potential buyer would have to shoulder the burden of addressing the contamination,” which probably created an independent basis for the jury to discount the property's value. Caffe I, 328 S.W.3d at 932. At retrial, the State exploited the exclusion of evidence concerning the State's role in delaying the property's remediation to create the same false impression: that Revere and Weatherford were not committed to cleaning up the property, and, consequently, that a potential buyer would have to shoulder the burden of addressing the contamination.

III. Conclusion

The trial court abused its discretion in excluding evidence concerning the State's role in delaying the condemned property's environmental cleanup, which went directly to the property's market value, and the court of appeals erred in affirming the exclusion. Accordingly, we reverse the court of appeals' judgment and remand the case to the trial court for a new trial in accordance with this opinion.