MAJORITY OPINION
SCOTT BRISTER, Chief Justice.Among the hundreds of wells drilled in the Tomball Field northwest of Houston in the 1930s, one was found to be leaking oil and gas into an underground aquifer in 1939. Humble Oil & Refining Company attempted to recover the lost hydrocarbons by backflowing” the well for almost two years.
Forty years later and a half mile to the southeast, a water well was drilled for the Three Lakes subdivision. When the well was first tested for organic chemical contaminants in 1990 (after about 10 years of use), it was discovered to contain benzene. The subdivision’s water supply was shifted to an uncontaminated source shortly thereafter.
In 1993, several hundred residents of Three Lakes sued Exxon Corporation, alleging the oilfield leak more than 50 years before involving its predecessor contaminated the water well and caused them various health problems. Six residents were selected for a bellwether trial, which took place over the better part of six weeks.
At the conclusion, the jury found Exxon negligent and grossly negligent, and awarded a total of almost $7 million in actual and punitive damages. In post-judgment proceedings, the trial court reduced or eliminated most of the jury awards, rendering take-nothing judgments against the four adults and reduced judgments for the two minors.
All parties appeal. Exxon contends there is no evidence its negligence caused either the contamination of the water well or the plaintiffs’ specific diseases. The plaintiffs contend the trial court erred in disregarding the jury’s awards. Both are governed by the same standard — whether any evidence supports the jury’s verdict.1 *180Because legal standards for reviewing medical causation are much clearer than those for underground hydrology, we address it first. Finding it dispositive, we go no further.
I. What We Review
Exxon challenges the opinions of the plaintiffs’ four medical experts, arguing they presented no scientific basis for concluding exposure to the water at Three Lakes caused the plaintiffs’ diseases. We consider all the evidence in the light most favorable to the verdict, and indulge every reasonable inference in that direction.2 But we do not take expert testimony at face value; if it is unreliable, it constitutes no evidence, and we must reverse.3
We disagree with the plaintiffs that two different approaches might apply — one based on the six factors applied in Merrell Dow Pharmaceuticals, Inc. v. Havner;4 and one that looks only for “analytical gaps” in the testimony.5 Undoubtedly, the tools used to test the reliability of expert testimony will vary depending on the field of expertise involved. But it is impossible to ignore the Havner factors here, as the field of expertise is the same — the epidemiological evidence connecting a chemical exposure and a disease.6
We also disagree with our dissenting colleague that Exxon waived error by failing to obtain a record from a pretrial hearing at which its motions to exclude allegedly unreliable expert testimony were first considered. Following this argument to its logical conclusion, every time a pretrial motion to strike expert testimony is denied without a reporter’s record, we must presume any jury verdict thereafter is valid (no matter how scientifically unsound). This argument is neither raised nor briefed by the parties, apparently for the same reason they never requested that the hearing be recorded — because no testimony was taken at the pretrial hearing. Not a single line in any party’s brief suggests we should look to the pretrial hearing for evidence of reliability, or that it included anything other than what was repeated a few days later at trial.
While this appears to be an issue of first impression, the Supreme Court has given us some guidance. First, it was the plaintiffs’ burden to present scientifically reliable evidence of causation at trial.7 Second, Exxon was required to object at trial (which it did repeatedly)8 so the *181plaintiffs would have an opportunity to cure any defects regarding reliability and present us with a fully developed record.9 Neither of these guidelines suggests a rule that, after its objections were overruled, it remained Exxon’s duty to incorporate into the trial record any and all pretrial evidence that supported the reliability of the plaintiffs ’ experts.10
It is generally true that unless an appellant files a complete reporter’s record (or a limited appeal11), we presume the omitted portions are relevant and support the jury’s verdict.12 But a complete record does not include matters from other proceedings; we normally test the legal sufficiency of a jury verdict by the evidence at the jury trial, not by what happened before or after it.13 Moreover, there is an important difference between presuming and pretending. Here, the trial record makes clear that no evidence was presented at the pretrial hearing.14 And if (as discussed below) the evidence at trial showed that scientific evidence could not support a particular verdict, any presumption that science held the opposite view a few days earlier has been effectively rebutted.15
Finally, our colleague’s reliance on a treatise discussing the admissibility of expert testimony16 shows he misconstrues *182Exxon's point. Challenges to both the admissibility and the legal sufficiency of expert testimony involve the same substantive guidelines for determining relevance and reliability.17 But admissibility of expert testimony is an initial decision made by the trial judge, usually outside the jury’s presence, and is subject to an abuse-of-discretion review.18 By contrast, legal sufficiency of expert testimony tests the jury’s decision after the trial, looking at the whole trial record to see if any evidence supports the verdict.19 Had Exxon complained of the pretrial ruling on admissibility, we would need a record of that hearing to review it. But Exxon does not. It is not our place to change this legal sufficiency point into something else, or dismiss it by reviewing a different decision by a different decision-maker in a different proceeding according to a different standard of review.
II. Benzene and ALL
James Makofski, Jr., was an infant when his family moved to Three Lakes in 1985. In September 1991, he was diagnosed with acute lymphocytic leukemia (ALL).20 Treatment of ALL is expensive and painful; in James’s case, it included three years of repeated rounds of chemotherapy and spinal injections. But the trial testimony also indicated current treatment is usually successful; in James’s case, annual checkups since 1994 have shown no indications of recurrence.
The plaintiffs alleged James’s ALL was a result of his exposure to benzene in the Three Lakes water supply. It was undisputed that current scientific knowledge cannot prove what causes ALL in any particular instance, and that in 95 percent of ALL cases the cause is never known. Because of so many uncertainties, efforts to establish liability for ALL based on exposure to hazardous substances have proved daunting.21 Here, the plaintiffs’ attorneys used epidemiological studies, a government study, and expert testimony to try to raise a fact question on causation.
The jury found Exxon’s negligence caused James’s injuries, and awarded him $1.5 million for past medical care, $1 million for future medical care, $400,000 for pain and anguish, and $325,000 for impairment. After reducing the award for future medical expenses to $156,603 (the only estimate of costs presented at trial), the trial court entered judgment in accordance with these awards. Exxon challenges all awards for lack of causation; the plaintiffs challenge the trial court’s reduction of the future medical award.
The Epidemiological Studies
In Havner, the Texas Supreme Court held that scientifically reliable epidemiological studies showing more than a doubling of the risk may establish a causa*183tion link.22 To benefit from this exception to the normal rules requiring individual proof of causation, a claimant must show a similar exposure to the subjects studied and exclude other plausible causes with reasonable certainty.23
Benzene has been known to be potentially harmful to humans for more than a century. Thus, a large number of epidemiological studies have been conducted on its effects. The parties’ experts agreed that benzene has been shown to cause acute myelogenous leukemia (AML), the most common form of leukemia found in adults. But the same studies have not reached the same conclusion regarding ALL, the most common form found in children.
Unfortunately, almost none of the relevant epidemiological studies appear in the trial court record, though several were marked for identification and discussed at length during the trial. As learned treatises, excerpts from these studies could be read to jurors, but none were admitted as exhibits for jurors’ independent review.24 But while the rules of evidence withhold learned treatises from jurors, that does not mean they should be withheld entirely from the record. Without them, we are hard-pressed to conduct the kind of review Havner requires.
Each party here says the other had the responsibility to place relevant studies in the record. A similar disagreement appears to exist among Texas appellate courts: the Sixth Court of Appeals has held studies cannot be considered on appeal unless they are “admitted into evidence,” 25 but the Fourth Court of Appeals has relied on studies attached to an appellate brief.26
In Havner, the Court discussed studies included in the record as well as studies that were only referred to in testimony. The Court found legally insufficient any studies as to which there was no information in the record (either from testimony or the study itself) establishing confidence levels and intervals.27 This example suggests testimony from a study may alone be enough, but it must be quite detailed.
Because of the uncertainty as to what sources we can review, we asked each party (over Exxon’s objection) to submit an appendix with copies of the studies on which they rely. Having reviewed them all, we find that even under the more lenient approach (allowing these studies to be supplemented on appeal) the evidence is legally insufficient under the Havner standards:
1. The Chinese Studies.28 This series of studies included more than 100,000 workers, the majority of whom had been exposed to benzene in factories in China. For ALL, the study found a relative risk of 2.8 with a confidence interval of 0.5 to *18454.5 at a 95% confidence level. As this interval includes 1.0, the finding was not statistically significant (that is, it was inconclusive).29 The authors concluded that risks were “markedly increased” for ALL, but that “the excess risk for this condition was not statistically significant.”30
2. The US/UK Study.31 This meta-analysis of more than 200,000 petroleum workers in the United States and United Kingdom found a relative risk for ALL of 1.32 with a confidence interval of 0.81 to 2.01 at a 95% confidence level. As this interval again includes 1.0, the finding was not statistically significant. The authors concluded that no increased risk for ALL or several other specific types of leukemia had been shown among petroleum workers in the two countries.32
3. The Port Arthur Study.33 This study looked at the causes of death among 18,000 workers at a refinery in Port Arthur, Texas between 1937 and 1978. For ALL, the study found a relative risk of 2.6 with a confidence interval of 1.1 to 5.1 at a 95% confidence level.34 But the authors concluded the ALL finding was spurious because of historical difficulties in distinguishing it from other leukemias:
From a diagnostic perspective, it might have been difficult historically to distinguish between patients with CLL [chronic lymphocytic leukemia] and those with ALL. This difficulty would likely have prevailed during a large portion of the cohort follow-up period.... Indeed, if even one ALL case was the result of misdiagnosis, the ALL [finding] would not have been statistically significant. In this regard, it is noteworthy that one of the ALL death certificates listed the cause of death as “Alymphatic [sic] leukemia”. As this is not a generally recognized diagnosis, we could not determine the cell type of this cancer. Nevertheless, we coded it as acute lym-phocytic leukemia in our analysis and as such, the statistical significance of the ALL result rests on a questionable diagnosis.... No epidemiologic study has demonstrated an association between exposure to benzene and acute lympho-cytic leukemia. Similarly, no other petroleum refinery study has reported an increase of ALL.... Thus, the finding from the Port Arthur refinery was inconsistent with other studies of refinery workers with similar exposures. The likely spurious nature of the ALL finding in the Port Arthur cohort is further supported by the lack of an exposure-response relationship based on length-of-employment analysis.35
4.Other Studies Submitted. Several other studies were submitted that contained no statistical information on any association between benzene and ALL. For example, one study found a significant association between ALL and other lymphatic leukemias with exposure to benzene and other solvents, but because of the *185confounding chemicals and diseases the study tells us nothing definitive about benzene and ALL.36 Another article stated it would be “surprising” if benzene caused only AML, but limitations in the literature precluded any more definitive conclusion.37
5. Other Studies Mentioned. The various experts briefly mentioned several other studies, but only in passing. For example, some testified that an author or a study “says benzene causes ALL” or “supports my opinion.” But none of these studies have been supplied to us (even after our specific request), and nothing gives us any indication of the scope, design, relative risk, confidence interval, confidence level, or even whether benzene or ALL were actually involved. Following Havner’s example, we cannot grant even a scintilla of weight to these passing references.38
The Three Lakes Registry
In addition to general epidemiological studies, the plaintiffs rely on a study by the Agency for Toxic Substances and Disease Registry (ATSDR), a division of the federal Department of Health and Human Services that gathers health information on persons exposed to hazardous substances. In 1989, the agency selected benzene to study because of its danger to humans, its ubiquity in the environment (from work sites, waste dumps, gasoline, automobile exhaust, tobacco products, and foods), and the lack of scientific information on the effect of low-level, long-term exposure.
In 1990, ATSDR chose to study the Three Lakes subdivision shortly after the contamination of the water supply was discovered and discontinued. Interviews were conducted in person and by telephone with about 1100 residents, with follow-up interviews in subsequent years. The data was collected in a “Benzene Su-bregistry” report, and in 1998 two of the primary investigators published a peer-reviewed study of the findings.39
In both, statistically significant increases at a 99% confidence level were reported (at least for some age groups in some years) for anemia, ulcers, gall bladder trouble, stomach or intestinal problems, stroke, urinary tract disorders, skin rashes, diabetes, kidney diseases, and respiratory allergies. By contrast, statistically significant decreases were found at the same confidence level for asthma, emphysema, arthritis, rheumatism, and hearing and speech impairments. No statistically significant differences were found for liver disease, mental retardation, or cancer. There was no finding whatsoever regarding ALL.
For several reasons, the ATSDR findings are insufficient to establish causation under the Havner standards. First and foremost, that is because the report itself says they cannot:
The findings in this report cannot be used to identify a causal relationship between the health outcomes and benzene exposure. In addition, there are some methodological differences in data collection that may have biased the re*186porting rates, resulting in false positive findings. The findings of this report do, however, reinforce the need to continue regular followup of this population.
Given this conclusion, the plaintiffs cannot use the subregistry data to establish causation. Havner does not allow us to give any weight to in-court conclusions about a study that disagree so diametrically with the published conclusions of the researchers who conducted it.40
Second, the ATSDR’s reference to potential bias recognizes the difficulty in gathering accurate data when many of the respondents are parties in this litigation. Experts for all parties agreed that those involved in litigation are likely to report higher levels of medical problems. The ATSDR researchers attempted to mitigate this problem by asking residents not whether they had a particular health condition, but whether a physician had told them they had it. What is missing from the plaintiffs’ efforts to use the ATSDR data is any scientific evidence of whether that mitigation was too little or too much.
Third, the ATSDR also noted another potential source for false positives — the large sample size and number of comparisons in the study. This comment apparently refers to the unfortunately-named “Texas Sharpshooter Fallacy,” in which natives of this state are alleged to shoot at the side of a bam and then draw a target where the most holes are located, thereby establishing the accuracy of their marksmanship.41 Epidemiologists use the term to identify the phenomenon that when a large number of health effects are surveyed, there is an increased likelihood that random chance will produce a statistically significant association when in fact none exists.42
Fourth, although the authors concluded the Three Lakes data could identify potential associations,43 that does not justify making causation conclusions from it. The ATSDR’s express purpose for studying Three Lakes was the lack of epidemiological information on the health effects of long-term exposure to low levels of benzene. As Havner notes, even statistically significant associations in an isolated study are no evidence of causation, as scientific methodology requires some confirmation.44
Finally, the ATSDR data is too general to support the kind of specific uses the plaintiffs try to make of it. In their brief, the plaintiffs make the rather shocking assertion that Exxon did not dispute general causation, despite days of cross-examination and counter witnesses called entirely for that purpose. Apparently, this is a reference to Exxon’s concession that benzene is associated with one type of leukemia (AML), but Exxon never conceded the same as to ALL. While lumping distinct diseases together as “leukemia” may yield a statistical increase as to the whole category, it does so only by ignoring proof that some types of the disease have a much greater association with benzene than others. The ATSDR made no attempt to distinguish among many particular types of diseases with apparently *187different etiologies. The categories of diseases the ATSDR uses — such as “anemia or other blood disorders,” “skin rashes, eczema, and other skin allergies,” or “cancers” — are simply too general to meet Havner’s standards.
The Experts
Finally, the plaintiffs also presented two experts to establish a causal connection between benzene and ALL: Dr. Bernard Goldstein, a physician board-certified in internal medicine, hematology, and toxicology, and Dr. Stuart Shalat, a doctor of epidemiology. Together, their testimony illustrates most of the practices Havner declares to be inconsistent with sound methodology among epidemiologists.
At trial, both experts testified that benzene causes ALL, despite the many published, peer-reviewed epidemiological studies finding no statistically significant relationship. Neither, however, appears to have ever offered the same opinion in an established scientific journal. In his own textbook, Dr. Goldstein listed the association between benzene and ALL as “suggested but unproven” in 1988, as “suggested” in 1986, and in 1998 as a “biomedical plausibility.”45 Plausibility is not enough.46 Havner instructs us to be especially skeptical of scientific evidence that has not been published or subjected to peer review.47
Dr. Goldstein explained his divergence from the published literature on his belief that a lower confidence level should be used in court. Although he admitted causation “from a scientifically proven point of view” requires 95 percent certainty, he imposed only a 51 percent certainty when testifying in court. Havner expressly rejects that argument:
The generally accepted significance level or confidence level in epidemiological studies is 95% ... Virtually all the published, peer-reviewed studies on Bendec-tin have a confidence level of at least 95%.... We think it unwise to depart from the methodology that is at present generally accepted among epidemiologists.48
Dr. Goldstein justified his departure from current epidemiological standards by pointing to lower standards used by OSHA, the EPA, or physicians who treat patients:
[Plaintiffs’ counsel]: If you wait till you have this 10-to-l exactitude that [Exxon is] urging, what’s going to really happen in this world?
[Dr. Goldstein]: There will be a lot of sick people ... I just wouldn’t be practicing good medicine or public health if I waited till I was absolutely certain.... I can’t wait as a physician until I’m 95 percent sure of something before I can deal with the patient’s potential problems.
Again, the Havner Court expressly distinguished between Type I (false positive) and Type II (false negative) statistical errors when deciding what confidence level is appropriate.49 As noted in the examples given in Havner, differences in costs and benefits make false positives acceptable in some situations but unacceptable in oth*188ers.50 For example, it may be appropriate for the EPA to protect people from chemical exposure on weak evidence that it will cause any harm, but that does not make it equally appropriate to impose a judgment of several million dollars on weak evidence that a defendant caused any harm. Hav-ner specifically bars this effort to lower epidemiological confidence levels for courtroom use.51
Additionally, Dr. Goldstein suggested causation conclusions regarding AML should lead to the same conclusions regarding ALL, as the diseases have “similar characteristics.” But he admitted all epidemiological studies in the last twenty years invariably treat these malignancies separately, and every study filed with us shows significantly different findings as to each. We cannot treat an epidemiological study regarding one disease as if it applied to another.52
Finally, Dr. Shalat reinterpreted the ATSDR’s inconclusive finding regarding cancer into a statistically significant one by adding to the study — which specifically did not study any deaths in the Three Lakes subdivision — statistics for the general public regarding deaths from cancer. Havner expressly prohibits reanalysis of a study to derive a significant finding if sound scientific methodology would not draw the same inference.53 Even if the incidence of all types of cancer was relevant (which it is not for the reasons stated above), there is no evidence that Dr. Shalat’s reconfiguration to reach a conclusion that the authors did not is scientifically sound.
Conclusion
In sum, the plaintiffs presented evidence that several experts believe a causal connection exists between benzene and ALL, while others do not. But an expert’s belief that a substance might cause a disease is no evidence that in reasonable probability it did.54 No epidemiological study establishes a statistically significant doubling of the risk of ALL from exposure to benzene; accordingly there was no evidence that— more probably than not — James’s ALL was caused by exposure to benzene in the Three Lakes water supply. We hold the evidence is legally insufficient to support James Makofski’s damage awards.
III. Benzene and Other Diseases
The only other plaintiff who obtained judgment against Exxon was John Russell. Born in November 1990, he was found to be anemic in a blood test taken five months after birth. Subsequent blood tests were occasionally “borderline,” but there has been no diagnosis of anemia since his early infancy. His parents testified he has suffered from recurrent nosebleeds, asthma, and skin rashes.
The jurors found John had suffered “$0” for all damages alleged in the past, but awarded $200,000 for future pain and mental anguish, $100,000 for future impairment, and $176,000 for future medical care. Exxon asserts there is no evidence John’s exposure to benzene caused any of his ailments, or that this exposure more than a decade ago will in reasonable probability result in any future damages.
*189As with leukemia, experts for both parties agreed that some types of anemia have been shown to be associated with benzene exposure and other types have not. Dr. Michael Gray, an internist board certified in occupational medicine who testified for the plaintiffs, said he believed John’s anemia was attributable to benzene, but admitted the medical records contained no studies excluding iron-deficiency anemia — the most common form, and one not associated with benzene. As John has not been diagnosed with anemia since then, there has been nothing to support this conclusion thereafter. Without establishing the type of John’s anemia, Dr. Gray did not exclude other plausible causes of it with reasonable certainty.55
Nor was there any other reliable scientific evidence connecting benzene to any of John’s other complaints. Dr. Gray admitted asthma was not caused by benzene (and as noted above, the ATSDR reported a statistically significant decrease in asthma at Three Lakes). He also testified benzene had suppressed John’s immune system, but offered no epidemiological studies or other scientific evidence connecting benzene exposure at levels John experienced with an appearance of nosebleeds, asthma, and skin rashes more than a decade later (as would be necessary to support the jurors’ awards of future damages only). Although nosebleeds, respiratory infections, and skin rashes have many causes, there is nothing in the trial record to exclude those alternate explanations but Dr. Gray’s naked assurance.
Finally, none of this evidence explains why the jury awarded nothing to John for past damages but a substantial amount for damages in the future. Clearly, the award was based on no disease he has exhibited to date. Instead, throughout the trial the plaintiffs’ counsel pointed to symptoms John had rather than diseases, and argued these symptoms were similar to those that preceded James Makofski’s ALL diagnosis. His entire closing argument for John’s damages was an attempt to connect the two boys’ ailments:
You look at John Russell.... What sum of money for John Russell? He’s eight years old. He’s at the age that James was. He’s got the same nosebleeds. He gets told about the same flus. He’s also got an asthma problem, not related in the sense that the asthma is caused by the benzene, but it makes every cold he has worse for the asthma. He’s got the anemia, as Dr. Gray told you, you know, this is a boy — you might check his anemia one day and it may be okay. Other days it’s not.... You won’t find many eight year olds that sit that still. That’s not just well-mannered. That’s a boy who is a little different because his body does not have the energy and because he’s got some problems and Dr. Gray’s testified to them. At this point is it leukemia? No.
Jurors obviously followed the unstated conclusion (“but he might get it in the future”) when they awarded no past damages but substantial future ones.
But not a single expert at trial supported this analogy. All agreed John did not have ALL, and none could say he probably ever would. His testifying expert admitted that no one could state with reasonable medical certainty that John would develop any form of cancer in the future, much less ALL. Moreover, as Exxon points out, the analogy with Makofski was misleading, as there was no evidence the latter had anemia or nosebleeds for several years before his ALL diagnosis. To the contrary, the only testimony at trial was that Makofski suffered nosebleeds and *190anemia immediately before his ALL diagnosis, due to the massive destruction of bone marrow cells that the disease engenders.
Texas law prohibits recovery for future diseases unless there is a reasonable medical probability the disease will occur.56 It also prohibits recovery of mental anguish damages for an increased risk of developing a disease that is not presently manifest.57 We find there was no legally sufficient evidence to support a causation connection between benzene exposure and any disease John Russell in reasonable probability will suffer in the future.
IV. Benzene and Mental Anguish
The remaining four adult plaintiffs asserted a wide range of medical problems continuing up to the time of trial, though by that time none of them had lived in Three Lakes for several years, or been exposed to benzene in the subdivision’s water supply for a decade. Andrea Russell (John’s mother) presented evidence that she suffers from anemia and keratos-es (benign precancerous skin lesions). Phillip Devora complained of skin rashes, acne, and fungal infections on his fingernails. His wife Janie sought damages for anemia, benign fibroid growths in her uterus, and excessive menstrual bleeding (which led to her decision to undergo a hysterectomy). Codi Stennett presented evidence she suffers from skin rashes, leukoplakia (depig-mentation of the skin), fungal infections, bronchial infections, and anemia. Several of these plaintiffs attributed some or all of their complaints to “immune suppression” caused by exposure to benzene.
Jurors apparently attributed almost none of these complaints to the water at Three Lakes, awarding “$0” as to every claim for damages except mental anguish (which they awarded in amounts ranging from $50,000 to $100,000). The plaintiffs do not challenge the jury’s verdict that they have suffered nothing but mental anguish. But they do challenge the trial court’s order setting aside the mental anguish awards.
Dr. John Wilson, an expert in post-traumatic stress syndrome (“PTSD”), admitted that while none of the four adults would meet current medical standards for that diagnosis, they nevertheless showed some or all of its symptoms. The plaintiffs argue the verdict could not have been based merely on their fear of contracting future diseases, as the trial court excluded such evidence. While it is true the trial court made that order (over the plaintiffs’ repeated objections), the record shows the evidence from the plaintiffs’ testifying psychologist was nothing but:
[As to Janie Devora]: She began to have worries about her family. She had worries about was she going to get a blood disorder. She began to have sleep disturbance. She began to express more stress and anxiety reactions. She began to have concerns learning that other people in the community had cancer.
[As to Phillip Devora]: He was a little bit what I would term guarded in his presentation; but underneath that, he also expressed, again, specifically, fears and anxieties and apprehensions about what was happening to him, what was happening to his wife and his family.... But his anxiety was apparent.
[As to Codi Stennett]: I think that the most succinct way I can put it is that she carries around this anxiety that *191she knows she’s been exposed to the benzene, that it’s had an effect on her, that it affects her cells and that she’s going to carry that with her.
[As to Andrea Russell]: She has anxiety. She has depression. She’s traumatized by this experience, and she has current concerns about her physical health and where that’s going to go starting now and in — from there on.
He agreed with the plaintiffs’ counsel that each of the plaintiffs suffered mental anguish because “they’ve been poisoned,” and while he admitted their experiences fell short of accepted PTSD causes such as combat or concentration camps, there was a “possible parallel” with victims of the latter because the plaintiffs here had been “gassed and survived.”
Clearly, there was substantial evidence the plaintiffs were anguished about their exposure to benzene in the Three Lakes water supply. But all of that evidence related to their fears about what it might mean for their health. Adopting the vivid metaphor used by plaintiffs’ counsel, people suffer mental anguish from drinking poison not because it tastes bad, but because they fear what it will do to them. The jury found these plaintiffs suffered no illness from their exposure to benzene at Three Lakes in the ten years since it stopped, or in reasonable probability will do so in the future. At this point so long after their exposure, it is clear that their anguish is less related to the water than to what they have been told about it.
Because Texas law does not permit recovery of mental anguish damages for fears related to developing a disease that has not occurred and has not been shown to be likely,58 we hold the trial court did not err in disregarding the plaintiffs’ mental anguish damage awards.59
Conclusion
Someday, medical science may find that benzene causes ALL; but on the record before us it is just as likely to find that it does not. Some of the other ailments the plaintiffs have suffered may also be connected to their exposure to benzene; but on the record before us we simply cannot tell. We cannot rush to impose liability when scientifically reliable evidence is unavailable.60
Accordingly, we reverse the judgment of the trial court, and render judgment that the plaintiffs take nothing.
SEYMORE, J., dissenting.
. See Merrell Dow Pharms., Inc. v. Havner, 953 S.W.2d 706, 711 (Tex.1997) (holding judgment in accordance with jury verdict improper if no scientifically reliable evidence *180supports it); Brown v. Bank of Galveston, N. A., 963 S.W.2d 511, 513 (Tex.1998) (holding judgment notwithstanding jury verdict proper only if no evidence supports it).
. See Havner, 953 S.W.2d at 711; see also Daubert v. Merrell Dow Pharms., Inc., 509 U.S. 579, 595, 113 S.Ct. 2786, 2798, 125 L.Ed.2d 469 (1993).
. See Havner, 953 S.W.2d at 714.
. Id.
. See Gammill v. Jack Williams Chevrolet, Inc., 972 S.W.2d 713, 727 (Tex.1998).
. See Daubert, 509 U.S. at 591-94, 113 S.Ct. at 2795-97; Havner, 953 S.W.2d at 711-14.
. See Havner, 953 S.W.2d at 720; E.I. du Pont de Nemours and Co. v. Robinson, 923 S.W.2d 549, 557 (Tex.1995).
. See Maritime Overseas Corp. v. Ellis, 971 S.W.2d 402, 409 (Tex.1998). Shortly before trial, Exxon filed a motion in limine requesting exclusion of several of the plaintiffs’ experts, including 60 pages of briefing, a 28-page appendix applying the brief to each plaintiff, 127 pages of deposition excerpts, and 55 pages of medical studies. Immediately before Dr. Goldstein and Dr. Shalat took the stand at trial, Exxon moved to exclude their testimony, and immediately after they finished moved to strike their testimony, each time requesting and the trial court allowing the motion in limine to be incorporated by reference in lieu of an extended oral objection.
. See Maritime Overseas, 971 S.W.2d at 412.
. We also disagree that this necessarily requires repetition of pretrial evidence before the jury. Either party at trial may incorporate by reference any proof from the pretrial hearing they wish made part of the trial record. Here, neither did.
. See TexR.Afp. P. 53(d).
. See Feldman v. Marks, 960 S.W.2d 613, 614 (Tex.1996).
. As neither party relies on any evidence from the pretrial hearing, we need not decide whether a legal sufficiency review should include evidence of scientific reliability that is never presented to jurors. See Campbell v. State, 85 S.W.3d 176, 184 (Tex.2002) (refusing to review issue of legal sufficiency without adequate briefing and argument); see also Harvey Brown, Procedural Issues Under Daubert, 36 Hous. L.Rev. 1133, 1139 (1999) (suggesting parties "presumably" need not re-offer evidence from pretrial hearing before jury, and that evidence from pretrial hearing will be considered "indirectly” in reviewing sufficiency of the evidence).
. At tried, Exxon's counsel introduced himself to each expert by noting previous depositions or hearings at which they had crossed swords, but no mention is made of a pretrial hearing a few days before trial. Each expert was examined at length about his deposition testimony, but none about testimony from any pretrial hearing. On the critical connection between benzene and leukemia, Dr. Bernard Goldberg admitted he provided neither animal studies nor epidemiological articles to Exxon’s counsel at any time between his deposition and trial. Finally, the very capable attorneys and experienced trial judge all certainly knew when a record is required and when it is not.
. The presumption that omitted portions of a record support a verdict is rebuttable. See Polanco v. Pan Amer. Univ., 818 S.W.2d 97, 99 (Tex.App.-Corpus Christi 1991, no writ); Candelier v. Ringstaff, 786 S.W.2d 41, 44 (Tex.App.-Beaumont 1990, writ denied); see also Garza v. Texas Alcoholic Beverage Comm'n, 83 S.W.3d 161, 166 (Tex.App.-El Paso 2000),(applying same rule to presumption in favor of judgments generally), aff'd, 89 S.W.3d 1 (Tex.2002) cf. Vickery v. Comm’n for Lawyer Discipline, 5 S.W.3d 241, 252, 256 (Tex.App.-Houston [14th Dist.] 1999, pet. denied) (holding denial of request for negative finding of fact contrary to court’s judgment did not rebut presumption of validity of trial court’s verdict). Thus, for example, Dr. Michael Gray’s trial testimony that there were no tests in John Russell’s medical records excluding iron-deficiency anemia (a form not associated with benzene) rebuts any presumption that the pretrial hearing included such evidence from the medical records.
. Michol O'Connor, O'CONNOR’S TEXAS RULES: [CIVIL TRIALS 289, § 3.5 (2002).]
. See Havner, 953 S.W.2d at 712.
. See Robinson, 923 S.W.2d at 558.
. See Havner, 953 S.W.2d at 720.
. Sometimes referred to as acute lymphob-lastic leukemia.
. See In re TMI Litigation, 193 F.3d 613 (3rd Cir.1999) (affirming summary judgment striking expert testimony on Daubert grounds that attempted to link ALL to radiation exposure from Three Mile Island incident); Whiting v. Boston Edison Co., 891 F.Supp. 12 (D.Mass.1995) (granting motion to strike expert testimony on Daubert grounds that attempted to link ALL to exposure to ionizing radiation); see also Jonathan Harr, A Civil Action (Random House 1995) (studying trial involving six cases of ALL allegedly caused by underground leakage of trichloroethylene into community water well in Woburn, Massachusetts).
. 953 S.W.2d at 718.
. Id. at 720.
. See TexR. Evid. 803(18).
. See Minn. Mining and Mfg. Co. v. Atterbury, 978 S.W.2d 183, 198 (Tex.App.-Texarkana 1998, pet. denied).
. See Texas Workers’Comp. Ins. Fund v. Lopez, 21 S.W.3d 358, 364-65 (Tex.App.-San Antonio 2000, pet. denied).
. See Havner, 953 S.W.2d at 725-26.
. Song-Nian Yin, et al., A Cohort Study of Cancer Among Benzene-exposed Workers in China: Overall Results, 29 Am. J. Indus. Med. 227 (1996); Song-Nian Yin, et al., An Expanded Cohort Study of Cancer Among Benzene-exposed Workers in China, 104 Envtl. Health Persp. 1339 (1996); Song-Nian Yin, et al., Leukaemia in Benzene Workers: a Retrospective Cohort Study, 44 Brit. J. Indus. Med. 124 (1987).
. See Havner, 953 S.W.2d at 723.
. Yin, Overall Results, 29 Am. J. Indus. Med. at 232, 233.
. Gerhard K. Raabe and Otto Wong, Leukemia Mortality by Cell Type in Petroleum Workers with Potential Exposure to Benzene, 104 Envtl. Health Persp. 1381 (1996).
. Id. at 1389.
. K.P. Satin, et al., A 50-year Mortality Follow-up of a Large Cohort of Oil Refinery Workers in Texas, 38 J. Occup. Envtl. Med. 492 (1996).
. For ease of comparison to the Havner standards, these figures have been converted from the standardized mortality ratios used in the article, which are based on a relative risk of 100.0 rather than 1.0.
.Satin, supra, at 503.
. Anthony J. McMichael, et al., Solvent Exposure and Leukemia among Rubber Workers: An Epidemiologic Study, 17 J. Occur. Med. 234 (1975).
. David A. Savitz and Kurds W. Andrews, Review of Epidemiologic Evidence on Benzene and Lymphatic and Hematopoietic Cancers, 31 Am. J. Indus. Med. 287 (1997).
. See Havner, 953 S.W.2d at 725-26.
. Jeanne R. Burg and Ginger L. Gist, The National Exposure Registry: Analyses of Health Outcomes From the Benzene Subregistry, 14 Toxicology & Indus. Health 367 (1998).
. 953 S.W.2d at 729-30 (holding expert's conclusion that substance caused birth defects, based on report in which researchers said it was only potentially capable of doing so, was not evidence of causation).
. See Boughton v. Cotter Corp., 65 F.3d 823, 835 n. 20 (10th Cir.1995); Robert F. Blom-quist, Bottomless Pit: Toxic Trials, The American Legal Profession, and Popular Perceptions of the Law, 81 Cornell L.Rev. 953, 960 n. 41 (1996) (book review).
. Id.
. Burg and Gist, supra, at 383.
. See Havner, 953 S.W.2d at 727.
. Dr. Goldstein testified he was upgrading this designation to "probable” in an edition not yet published, but did not indicate what (if any) peer-review was involved in the publication.
. See Havner, 953 S.W.2d at 729.
. Id. at 726-27.
. Id. at 723-24.
. Id. at 722-23.
. Id.
. Id. at 722-24.
. Id. at 725 (holding published studies showing statistically significant results regarding birth defects other than limb reduction defects could not support a finding regarding limb reduction defects).
. Id. at 720, 727.
. Id. at 729.
. Id. at 720.
. See Pustejovsky v. Rapid-American Corp., 35 S.W.3d 643, 652 (Tex.2000).
. See Temple-Inland Forest Prods. Corp. v. Carter, 993 S.W.2d 88, 93 (Tex.1999).
. See Carter, 993 S.W.2d at 93.
. Jurors also awarded each of the six plaintiffs $500,000 in punitive damages. As we have found there is no evidence supporting any of the plaintiffs' actual damage awards, we also set aside the awards of punitive damages. See Twin City Fire Ins. Co. v. Davis, 904 S.W.2d 663, 665 (Tex.1995).
.See Havner, 953 S.W.2d at 728.