OPINION

RICHARD H. EDELMAN, Justice.

Liana Smith ("Smith") appeals a take-nothing summary judgment entered in favor of Landry's Crab Shack, Inc. d/b/a Joe's Crab Shack ("Landry's") on the grounds that: (1) there is legally sufficient evidence that Smith suffered food poisoning from food she ate at a Landry's restaurant (the "restaurant"); and (2) her expert's testimony should not have been stricken. We affirm.

Smith's suit against Landry's alleged causes of action for strict liability, breach of warranties, deceptive trade practices, and negligence. Landry's filed a motion to strike Smith's expert causation evidence for lack of reliability and a motion for summary judgment, asserting that there was no cause in fact evidence that the food Smith ate at the restaurant caused her illness. The trial court granted both.

A no-evidence motion for summary judgment must be granted if: (1) the moving party asserts that there is no evidence of one or more specified elements of a claim or defense on which the adverse party would have the burden of proof at trial; and (2) the respondent produces no summary judgment evidence raising a genuine issue of material fact on those elements. See TEX.R. CIV. P. 166a(i). In reviewing a no-evidence motion for summary judgment, we view all of the summary judgment evidence in the light most favorable to the nonmovant, indulging every reasonable inference and resolving any doubts against the motion. City of Keller v. Wilson, 168 S.W.3d 802, 823-25 (Tex. 2005).

As applied to this case, the test for cause in fact is whether the illness would not have occurred if Smith had not eaten at the restaurant. See Marathon Corp. v. Pitzner, 106 S.W.3d 724, 727 (Tex. 2003). A finding of cause in fact may be based on circumstantial evidence, but cannot be supported by mere conjecture, guess, or speculation. Id. Similarly, a fact may not be inferred from circumstantial evidence that could support multiple inferences if none are more probable than the others. Lozano v. Lozano, 52 S.W.3d 141, 148 (Tex. 2001). In addition, an inference drawn only from other inferences (rather than direct evidence) is not legally sufficient evidence. Marathon Corp., 106 S.W.3d at 728.

To constitute evidence of medical causation, an expert opinion must rest on reasonable medical probability. Burroughs Wellcome Co. v. Crye, 907 S.W.2d 497, 500 (Tex. 1995). Reasonable probability is determined by the substance and context of an expert opinion, rather than the use of any particular words. Id. Thus, where the substance of an expert's testimony establishes only a mere possibility, rather than a reasonable probability, of causation, it is no evidence of causation. See Schaefer v. Tex. Employers' Ins. Ass'n, 612 S.W.2d 199, 204-05 (Tex. 1980).

Smith's first issue contends that a fact issue was raised on cause in fact by the testimony of her medical expert witness, Dr. Al-Assi, that Smith had food poisoning and, based on the incubation period, it was caused by the meal. We do not agree.

Smith provided testimony that: (1) the food she ate at the restaurant was the only food she had consumed within the 20 hours preceding the onset of her illness; and (2) the meal she had eaten prior to the one at the restaurant had also been eaten by her entire family without incident. Her medical expert witness, Dr. Al-Assi, testified that: (1) none of the circumstances of Smith's illness or treatment was inconsistent with a diagnosis of gastroenteritis; (2) gastroenteritis can be caused by bacteria in food; (3) many foods can be a carrier for such bacteria; (4) different bacteria have different incubation periods before a person begins to experience symptoms; (5) starting to experience symptoms within five hours of a meal and getting sick enough to be admitted to a hospital within about ten hours of a meal would, in reasonable medical probability, be "consistent with" food-borne gastroenteritis; (6) the "most common" time period in which symptoms of food-borne gastroenteritis "normally" manifest is within the first 24 hours and it can be as fast as a couple of hours; and (7) although it was possible that Smith had gastroenteritis from some other source than food poisoning, it was "more likely" that she had food poisoning considering "the clinical history and the set up."

However, Al-Assi did not express any ultimate opinion (in reasonable medical probability or otherwise) that Smith's gastroenteritis was caused by food she ate at the restaurant. Moreover, because his opinion that Smith's gastroenteritis was caused by food rather than another source was not stated as being in reasonable medical probability, but only as being more likely, and because he provided no factual basis or explanation for why the clinical history and set up showed this to be the case, this opinion was not sufficient to raise a fact issue. Therefore, Smith provided evidence only that she had gastroenteritis that could have been caused by food and that the only food she had eaten before experiencing those symptoms and within the period in which food poisoning symptoms most commonly occur was the meal she ate at the restaurant. Without inferring: (1) that Smith's gastroenteritis was caused by food; and either that (2) it occurred within the maximum (versus most common) incubation period for food-borne gastroenteritis; or (3) it was a type of food-borne gastroenteritis that occurs within the most common incubation period, none of which are supported by evidence in this case, it does not logically follow that her illness was caused by food she ate at the restaurant. Under these circumstances, the evidence does not raise a fact issue whether Smith's illness was, in reasonable medical probability, caused by food she ate at the restaurant. Therefore, Smith's issues are overruled, and the judgment of the trial court is affirmed.

SEYMORE, J. concurring in result only.