The issue presented by this appeal is how a court should determine the admissibility of expert opinion testimony. At the summary judgment stage of this ease, plaintiff attempted to establish medical causation of a toxic tort through the testimony of a single expert witness. The district court held that the basis of the expert’s opinion was insufficiently reliable and, in the alternative, that the expert’s testimony would have been more prejudicial than probative. With the expert’s testimony ruled inadmissible, plaintiff was left without proof of causation. The district court entered summary judgment for the defendants. We affirm.
I.
Christophersen died in March of 1986 as a result of a rare, small-cell form of cancer that originated in his colon and metastasized to his liver. During the fourteen years preceding his death, Christophersen worked for Marathon at its plant in Waco, Texas. At that plant, Marathon produces nickel/cadmium batteries. Christophersen never was directly involved in the production of these batteries. The record, however, indicates that over a number of years Christophersen’s job duties required him to visit the area of the plant in which the batteries were manufactured. During these visits, Christophersen was allegedly exposed to fumes resulting from the manufacturing process. Plaintiffs, Christopher-sen’s surviving spouse and child, contend that these fumes contained particles of nickel and cadmium and that Christopher-sen’s exposure to these heavy metals caused the cancer that resulted in his death.
Plaintiffs brought suit pursuant to the Texas Wrongful Death and Survival Statute, Tex.Civ.Prac. & Rem.Code Ann. §§ 71.001-031 (Vernon 1986), against Marathon and a number of companies that supplied Marathon with chemicals and other materials used in the manufacture of the nickel/cadmium batteries. Plaintiffs' complaint alleged that the products used in the production of the batteries were defectively designed, manufactured, and marketed, and were the producing causes of the cancer that resulted in Christophersen’s death. The complaint also alleged that Marathon was aware of the dangerous nature of the chemicals and products and failed to provide Christophersen a safe place to work or to warn him of the dangerous conditions *1109that existed at the plant. Marathon moved for summary judgment. The district court determined that plaintiffs failed to state a design defect or manufacturing defect claim and granted the defendants’ motions for summary judgment. Plaintiffs do not appeal this ruling.
The district court also granted Marathon’s motion for summary judgment on the marketing defect claim because the plaintiffs did not present sufficient evidence of causation. In reaching this conclusion, the court focused on the affidavit of the plaintiffs’ expert witness, Dr. Miller, who concluded that Christophersen’s exposure to nickel and cadmium at Marathon caused the cancer that resulted in his death. The district court undertook an in-depth review of the basis for Dr. Miller’s conclusion and determined that his opinion should be excluded. On appeal, a panel of this court reversed, holding that Dr. Miller’s opinion was not so fundamentally unreliable that the jury should not consider it. Defendants timely petitioned this court for rehearing en banc.
II.
A trial court’s ruling regarding admissibility of expert testimony is protected by an ambit of discretion and must be sustained unless manifestly erroneous.1 At the same time, we accord “proper deference to the jury’s role as the arbiter of disputes between conflicting opinions. As a general rule, questions relating to the bases and sources of an expert’s opinion affect the weight to be assigned that opinion rather than its admissibility and should be left for the jury’s consideration.” Viterbo v. Dow Chem., 826 F.2d 420, 422 (5th Cir.1987); see also Dixon v. International Harvester, 754 F.2d 573, 580 (5th Cir.1985).
This is an appeal from a grant of summary judgment. Although we review grants of summary judgment de novo, that is, under the same Rule 56 standards as are used by the district court, e.g., Medlin v. Palmer, 874 F.2d 1085, 1089 (5th Cir.1989) (applying the Celotex and Rule 56 summary judgment standards de novo) (citing United States Steel Corp. v. Darby, 516 F.2d 961 (5th Cir.1975)), in Rule 56 proceedings we still apply the manifest-error standard of review to the trial court’s evidentiary rulings, Lavespere v. Niagara Mach. & Tool Works, 910 F.2d 167, 175-76 (5th Cir.1990); Slaughter v. Southern Talc Co., 919 F.2d 304, 306-07 (5th Cir.1990); Washington v. Armstrong World Indus., 839 F.2d 1121, 1123 (5th Cir.1988); Viterbo, 826 F.2d at 422.2 Thus an appeal of a summary judgment presenting evidentiary issues raises two levels of inquiry. At the first level, we review the trial court’s evidentiary rulings, which define the summary judgment record, and we give these rulings their due deference. At the second level, with the record defined, we review the trial court’s summary judgment decision de novo. When the contested evidence is essential to the cause of action and the trial court has excluded the evidence, we may decide the appeal at the first level solely on the basis of the soundness of the evidentiary ruling. For if we uphold the exclusion of essential evidence, the second-level inquiry becomes academic. E.g., Viterbo, 826 F.2d at 422 (stating that appropriateness of summary judgment depends solely upon whether the district court erred in excluding causation testimony of plaintiff’s expert). This is such a case. With*1110out Dr. Miller's testimony, Christophersen cannot prove that any exposure to nickel/cadmium caused the colon cancer. In sum, we ask in this appeal whether excluding Dr. Miller’s opinion was manifestly erroneous.
III.
The Federal Rules of Evidence, combined with Frye v. United States, 293 F. 1013 (D.C.Cir.1923), provide a framework for trial judges struggling with proffered expert testimony. The signals are not neatly cabined categories, and we disentangle them only to accent the independent significance of each.
(1) Whether the witness is qualified to express an expert opinion, Fed.R.Evid. 702;
(2) whether the facts upon which the expert relies are the same type as are relied upon by other experts in the field, Fed.R.Evid. 703;
(3) whether in reaching his conclusion the expert used a well-founded methodology, Frye; and
(4) assuming the expert's testimony has passed Rules 702 and 703, and the Frye test, whether under Fed.R.Evid. 403 the testimony’s potential for unfair prejudice substantially outweighs its probative value.
These four signals or inquiries introduce no new concepts to our jurisprudence. They are only guideposts drawn from the Federal Rules of Evidence and our cases. We list these inquiries, but in doing so we do not intend that they be applied mechanically. At the same time, they often will naturally lend themselves to sequential application. The reality is that trials are too varied for fixed molds; we construct none today.
The first three steps are best understood as threshold requirements that all expert testimony must meet before being deemed admissible. Cf. Gideon v. Johns-Manville Sales Corp., 761 F.2d 1129, 1135 (5th Cir.1985) (construing Rule 702’s qualifications requirement as a threshold inquiry); Slaughter, 919 F.2d at 306-07 (construing Rule 703’s factual basis requirement as a threshold inquiry). Rule 403, on the other hand, provides an overlay—a final mechanism for screening out otherwise admissible testimony whose potential for prejudice substantially outweighs its probative value.3 Cf. 22 C. Wright & K. Graham, Federal Practice & Procedure, § 5213 at 258-59 (1978).
The first question concerns the expert’s qualifications: Is the witness—because of his specialized knowledge, skill, experience, training, or education in the relevant field—qualified to express an expert opinion on the topic at issue? Fed.R. Evid. 702. The Advisory Committee Note accompanying Rule 702 reads the broad language of the rule to permit expert testimony not only by experts carrying formal credentials such as university degrees and professional memberships but also by so-called skilled witnesses, whose experiences permit them to testify with authority on a given topic. The areas of inquiry that expert testimony may address are similarly broad, including scientific and technical questions as well as any other area of specialized knowledge. An expert may testify in his area of expertise “in the form of an opinion or otherwise.” This much is rote. The more subtle problem, and our caveat, is that the inquiry into the qualifications of an expert should not be a substitute for scrutinizing an expert’s reasoning or methodology. At this stage, the only question for the trial court is whether the expert is generally qualified to render an opinion on the question in issue.
Second, if the expert is qualified, are the facts and data that serve as a basis for the expert’s opinion the same type of facts as other experts in the same field reasonably rely upon in forming their opinions? Fed.R.Evid. 703.4 While testimony *1111based on the personal observations of the expert is preferable, neither the rules nor our cases have insisted on personal examinations.5 The reports and statements of others such as doctors, nurses, or medical personnel, while not as valuable as testimony based on the expert’s own observations, can provide a reliable basis for the expert’s opinion, at least when reliance on such sources is the custom of the discipline. At the same time, a common-sense skepticism may be warranted when an expert’s factual basis is derived, not from treatment or observation, but from subjective information obtained from counsel or client in preparation for trial.6 But such skepticism should not necessarily lead us to exclude the expert’s opinion. So long as the facts upon which the expert bases his opinion are those “perceived or made known to the expert at or before the hearing” and are “of a type reasonably relied upon by experts in the particular field,”7 we should proceed to evaluate the expert’s methodology. Fed.R.Evid. 703.
The third factor is the Frye test. We ask whether in reaching his conclusion, the witness used a well-founded methodology or mode of reasoning, one “sufficiently established to have gained general acceptance in the particular field in which it belongs.” Frye, 293 F. at 1014; see also Osburn v. Anchor Laboratories, Inc., 825 F.2d 908 (5th Cir.1987), cert. denied, 485 U.S. 1009, 108 S.Ct. 1476, 99 L.Ed.2d 705 (1988). As long as the expert’s methodology is well founded, the nature of the expert’s conclusion is generally irrelevant, even if it is controversial or unique. See Peteet v. Dow Chemical Co., 868 F.2d 1428 at 1433 (5th Cir.1989); Osburn, 825 F.2d at 915.8 In fact, in Osbum the plaintiff’s and the defendant’s experts relied on essentially the same diagnostic methodologies but drew opposite conclusions from the available information. We did not attempt to determine which expert’s conclusion was more in line with the consensus in the scientific community. Instead we stated, “a jury must be allowed to make credibility determinations and weigh the conflicting evidence in order to decide the likely truth of a matter not itself initially resolvable by common knowledge or lay reasoning.” Id. at 916. “An expert’s opinion need not be generally accepted in the scientific community before it can be sufficiently reliable and probative in support of a jury finding.” Osburn, 825 F.2d at 915 (emphasis added).9
*1112Finally, Rule 403 serves a general screening function for otherwise admissible evidence. Assuming the witness’s testimony is relevant and the witness is qualified, and assuming his testimony has the appropriate factual basis and his methodology is well founded, the only remaining inquiry is the balancing authorized in Rule 403: “Although relevant, evidence may be excluded if its probative value is substantially outweighed by the danger of unfair prejudice, confusion of the issues, or misleading the jury, or by considerations of undue delay, waste of time, or needless presentation of cumulative evidence.” In the 403 balancing inquiry, courts may consider issues that were appropriately excluded from the three preceding threshold inquiries, which are narrower in scope. If, taking all considerations into account, a court finds that the potential for prejudice substantially outweighs the probative value, the court may opt to exclude the testimony.10 Unlike Rules 702 and 703, which define admissibility thresholds for expert-opinion evidence, Rule 403 “creates a power to exclude otherwise admissible evidence; it gives no discretion to admit evidence that is subject to exclusion under some other rule.” 22 C. Wright & K. Graham, Federal Practice & Procedure, § 5213 at 258-59 (1978). We caution, however, that clearing the hurdles of Article VII of the Federal Rules of Evidence will not alone ensure approval under Rule 403. An application of these principles to this case follows.
IV.
Rule 702: Qualifications
Although the district court did not conclude that Dr. Miller was unqualified to testify under Rule 702, it did question his qualifications:
Dr. Miller is not an expert in either oncology or pathology. Miller’s opinion as to the cause of Christophersen’s death was formed without consultation with oncologists or other cancer specialists. Dr. Miller’s experience with cancer occurred during his residency when he assisted in a study of the immune system as affected by smoking and asbestos. Dr. Miller does not routinely treat cancer patients, nor has he ever treated a patient with a colon cancer of the type that affected Christophersen.
(citations omitted).
The trial court’s concerns are understandable. As we stated in Peteet, “district judges and appellate courts must carefully review an expert’s testimony to ensure that the expert has the necessary qualifications and a sufficient basis for his opinion.” 868 F.2d at 1431; Lavespere, 910 F.2d at 176 (citing Peteet); see also In Re Air Crash Disaster at New Orleans, Louisiana, 795 F.2d 1230 at 1234-35 (5th Cir.1986) (“Trial judges must be sensitive to the qualifications of persons claiming to be an expert.”). We caution, however, that although credentials can be significant, they alone are not necessarily determinative. The questions, for example, do not *1113stop if the expert has an M.D. degree. That alone is not enough to qualify him to give an opinion on every conceivable medical question. This is because the inquiry must be into actual qualification — sufficient to assist the trier of fact. Viterbo, 826 F.2d at 422. The trial judge here rightly scrutinized Dr. Miller’s lack of specialized experience and knowledge. The district court did not base its disallowance of Dr. Miller’s testimony on this issue, however. It preferred to rely on Rule 703 and the witnesses’ methodology. We, therefore, move to those issues.
Rule 703
Dr. Miller premised his opinion that Marathon caused Christophersen’s cancer on his belief that Christophersen had approximately a twenty-year history of “extensive exposure to nickel and cadmium fumes in the work place.” The district court, pursuant to Rule 703, analyzed the underlying “facts and data” of Dr. Miller’s opinion to determine whether it was based on the types of facts reasonably relied upon by experts in the field.11 Dr. Miller testified at his deposition that the level and duration of the patient’s exposure are important considerations when evaluating the effect of exposure to a toxic substance.
The district court found that virtually all of the factual data concerning Christopher-sen’s exposure to nickel and cadmium came from the affidavit of a Marathon employee named Edgar Manoliu (Manoliu), who described the fumes and Christophersen’s exposure to them. The district court criticized this affidavit, however, as being inaccurate and incomplete. The Manoliu affidavit appears to have over-estimated the number of times per week Christophersen visited the manufacturing area, as well as the average length of time he would remain there on each visit. The affidavit was also devoid of any information about the type of fumes to which Christophersen was exposed or the type of fumes generated by the battery manufacturing process.
We find particularly telling Manoliu’s admission in his deposition that he did not know the chemical composition of the fumes nor the mix of chemicals in the impregnation and soak tanks.
Nor was Dr. Miller informed as to the physical facilities at the Marathon plant, including the size of the plant or the impregnation and soak area, or the ventilation available in these areas or in Christopher-sen’s office. In addition, Dr. Miller did not always rely upon the accurate data that were contained in the affidavit. For example, the affidavit correctly indicated that Christophersen worked for fourteen years at the Waco plant before his death. Dr. Miller, however, based his opinion upon the assumption that Christophersen worked in the plant for twenty years. Thus Dr. Miller over-estimated the duration of Christo-phersen’s exposure by approximately fifty percent. Accordingly, accurate dosage and exposure information was not used by Dr. Miller.
Thus the court validly called into question the facts and data relied upon by Dr. Miller in forming his opinion. Slaughter v. Southern Talc Co., 919 F.2d 304 (5th Cir.1990), reiterated the requirements under Rule 703:
As a general rule, questions regarding the scientific bases of an expert’s opinion “affect the weight to be assigned that opinion rather than its admissibility and should be left for the jury’s consideration.” Viterbo v. Dow Chemical Co., 826 F.2d 420, 422 (5th Cir.1987). See also, Dixon v. International Harvester Co., 754 F.2d 573, 580 (5th Cir.1985). However, this general rule yields when “the source upon which an expert’s opinion relies is of such little weight ... that [the] testimony would not actually assist the jury in arriving at an intelligent and sound verdict.” Viterbo, 826 F.2d at *1114422. In such case, this court requires courts to examine the reliability of an expert’s sources to determine whether they satisfy the threshold established by the rule. See Soden v. Freightliner Corp., 714 F.2d 498, 502 (5th Cir.1983).
Id. at 306-07.
Plaintiffs do not contest the district court’s findings as to the deficiencies in the Manoliu affidavit. Rather, they argue that Dr. Miller stated in his opinion that dosage was less important when determining individual causation. Plaintiffs accordingly argue that any deficiencies in the underlying facts and data go to the weight of Dr. Miller’s opinion rather than its admissibility. We disagree. If the dosage of the harmful substance and the duration of exposure to it are the types of information upon which experts reasonably rely when forming opinions on the subject, then the district court was justified in excluding Dr. Miller’s opinion that is based upon critically incomplete or grossly inaccurate dosage or duration data. See Soden, 714 F.2d at 506. See also Slaughter v. Southern Talc Co., 919 F.2d 304, 306 (5th Cir.1990) (affidavits of two doctors that “represented nothing more than bare conclusions derived from erroneous data” held inadmissible for summary judgment purposes); Washington v. Armstrong World Indus., Inc., 839 F.2d 1121, 1123-24 (5th Cir.1988) (expert’s conclusions “lacked probative value because it was pure speculation based on negative inferences drawn from the testimony of three treating physicians”); Thompson v. Southern Pac. Transp. Co., 809 F.2d 1167 (5th Cir.1987), (expert’s opinion on dioxin as source of plaintiff’s illness had “insufficient factual basis” because expert had no knowledge of the amount or duration of the exposure).
As we have noted, Rule 703 seeks to ensure that the “facts and data” not otherwise admissible in evidence that form the basis of an expert’s opinion are “of a type reasonably relied upon by experts in the particular field in forming opinions or inferences upon the subject.” Although this rule is primarily directed toward permitting an expert to base his opinion on hearsay or otherwise inadmissible sources, Barrel of Fun, Inc. v. State Farm Fire & Casualty Co., 739 F.2d 1028, 1033 (5th Cir.1984), the inquiry into the “types” of “facts and data” underlying an expert's testimony is not limited to the admissibility of that data. District judges may reject opinions founded on critical facts that are plainly untrustworthy, principally because such an opinion cannot be helpful to the jury.12
The argument that Rule 703 addresses only generic facts and data and is unconcerned with the sufficiency and accuracy of underlying facts as they relate to the case at hand, will lead to the irrational result that Rule 703 requires the court to admit an expert’s opinion even if those facts and data upon which the opinion are based are crucially different from the undisputed record. Such an interpretation often will render Rule 703 impotent as a tool for testing the trustworthiness of the facts and data underlying the expert’s opinion in a given trial. Certainly nothing in Rule 703 requires a court to admit an opinion based on facts that are indisputably wrong. Even if Rule 703 will not require the exclusion of such an unfounded opinion, general principles of relevance will. In other words, an opinion based totally on incorrect facts will not speak to the case at hand and hence will be irrelevant. In any event such an opinion will not advance the express *1115goal of “assisting the trier of fact” under Rule 702.13
We do not of course say that Rule 703 requires that all facts and data underlying the opinion must relate perfectly to the record facts. As we have pointed out, only when the facts and data are critically inaccurate or incomplete, as determined by what other experts would or would not be willing to base opinions upon, would the facts and data lack the necessary requisites of Rule 703. The district court in this case did not abuse its discretion.
Frye: Well-Founded Methodology
When analyzing the validity of an expert’s methodology, we seek to determine whether it connects the facts to the conclusion in a scientifically valid way.14 We answer this question by applying the Frye test: whether the methodology or reasoning that the expert uses to connect the facts to his conclusion is generally accepted within the relevant scientific community.15
In his deposition Dr. Miller stated that the kinds of evidence most often used to establish causation are human epidemiological studies, live animal testing, and in vitro testing. Defendants’ experts agreed, but went a step further; they stated that the determination of the pathogenesis of a particular type of cancer requires clearly positive results from one or more of these types of testing. The district court was persuaded that it was inconsistent for Dr. Miller, on one hand, to conclude that these are the main methodologies and, on the other hand, to concede that he did not effectively rely on any of them. That epidemiological, animal, and in vitro studies are, as Dr. Miller said, the primary methods by which medical experts develop their theories of causation is not to say that Miller’s methodology was invalid. The Frye question focuses on the proffered methodology alone and looks to the scientific community to determine if general support for that methodology exists.
The critical portion of Dr. Miller’s opinion, as it relates to causation, is as follows:
‘the same sorts of chemicals and exposures that are associated with small-cell carcinoma of the lung are likely to be associated with small-cell carcinoma elsewhere in the body.’
Christophersen v. Allied-Signal Corp., 902 F.2d 362, 365-66 (5th Cir.1990). Dr. Miller offered no scientific methodology to support this assertion. Dr. Sherwood Gorbach concluded that:
Dr. Miller’s presumption that nickel and cadmium have been associated with a certain type of cell in lung cancer and therefore should be associated with a similar type of cell in the colon has no support in medical science and is without foundation.
The other defense expert, Dr. Richard Rudder, held the same view of Dr. Miller’s methodology.16 All Dr. Miller had was a scientific hunch, which as far as the record shows, no one else shares. This was enough to support further investigation but was inadequate to support a judgment in favor of Christophersen. The district court found that “Dr. Miller’s conclusion that a small cell carcinoma of the lung is *1116likely to be associated with a small cell carcinoma located elsewhere in the body is ‘without precedent in cancer epidemiology and is not scientifically correct.’ ” This finding of what is a scientifically correct conclusion is not for the district court. However, we cannot say that the district court erred in excluding Dr. Miller’s testimony. To the contrary, the district court was within its discretion in concluding, albeit implicitly, that Dr. Miller’s testimony failed to meet the third threshold test, the Frye test.
Rule 403: Prejudice versus Probativeness
Because we find that Dr. Miller’s testimony failed to clear either the Rule 703 or the Frye hurdle, we need not consider the district court’s application of Rule 403.
V. CONCLUSION
The foregoing analysis demonstrates that this was not a “battle of experts” that would be improper for resolution on summary judgment. The district court’s ruling that Dr. Miller’s opinion was inadmissible was not manifestly erroneous. Because Dr. Miller’s testimony was the only evidence of causation, the district court did not err in granting summary judgment for defendants.
AFFIRMED.
. See, e.g., Salem v. United States Lines Co., 370 U.S. 31, 35, 82 S.Ct. 1119, 1122, 8 L.Ed.2d 313 (1962); Spring Co. v. Edgar, 99 U.S. 645, 658, 25 L.Ed. 487 (1878); Myers v. Griffin-Alexander Drilling Co., 910 F.2d 1252, 1254 (5th Cir.1990); Washington v. Armstrong World Indus., 839 F.2d 1121, 1123 (5th Cir.1988); Viterbo v. Dow Chem., 826 F.2d 420, 422 (5th Cir.1987); Crawford v. North, 447 F.2d 738, 740-41 (5th Cir.1971).
. For the proposition that materials relied upon at a summary judgment hearing must contain only information that is admissible at trial, see generally Fed.R.Civ.P. 56(e) (discussing affidavits); C. Wright, A. Miller, & M. Kane, 10A Fed.Prac. & Proc. § 2722 at 48-50 & n. 12 (depositions), § 2738 at 470 & n. 7, 474 & n. 12 (affidavits) (2d ed. 1983). Cf. Liberty Leasing Co. v. Hillsum Sales Corp., 380 F.2d 1013, 1015 (5th Cir.1967) (upholding district court's decision to disregard deposition that contained inadmissible hearsay), cited in Samuels v. Doctors Hosp. Inc., 588 F.2d 485, 487 (5th Cir.1979).
. Of course, Rule 402’s requirement that, to be admissible, all evidence must be relevant applies with the same force to expert testimony as it does to all forms of evidence.
. Note that for purposes of Rule 703 this question is applicable only if the facts and data relied upon are not independently admissible. Fed.R.Evid. 703.
. Washington, 839 F.2d 1121.
. Viterbo, 826 F.2d at 423.
. See Greenwood Util. Comm’n. v. Mississippi Power Co., 751 F.2d 1484, 1495 (5th Cir.1985) (stating that, had trial court’s ultimate decision turned on expert’s testimony, appellate court would have remanded for factual finding whether the evidence relied on was of the type usually relied on by experts; also stating that trial court should accord deference to expert’s assertion that the experts in the field rely on similar evidence) (dicta).
. This principle, however, is not wholly open-ended, but, as we observed in Osbum, is rather designed for application in instances "where science has some meaningful information” to offer on the subject — as opposed to mere “theoretical speculation" — and "scientific ‘truth’ has not so completely hardened as to prevent legitimate difference of true expert opinion in a particular concrete context." Id. at 915 n. 10. If science has thus hardened or if what science offers is essentially no more than theoretical speculation, then well-founded methodology and reasoning may not alone suffice. See Brock v. Merrett Dow Pharmaceuticals, 874 F.2d 307 (5th Cir.1989), cert. denied, — U.S. —, 110 S.Ct. 1511, 108 L.Ed.2d 646 (1990).
Moreover, even though the Osbum criteria are met, the opinion may nevertheless be ex-cludable under Rule 403.
.One commentator has sought to capture the distinction between an expert’s methodology (or reasoning) and an expert’s conclusion by evaluating the former according to its scientific “validity” and the latter according to its legal "reliability.” Bert Black, A Unified Theory of Scientific Evidence, 56 Fordham L.Rev. 595 (1988). Black explains the distinction as follows;
Validity and reliability, though intertwined, are very different concepts. One normally speaks of "valid” rather than "reliable” reasons or theories, and of "reliable” rather than "valid" instruments or machines. Results, conclusions, or techniques may be either valid or reliable. Behind these simple examples of everyday usage lie largely overlooked conceptual distinctions and relationships that are fundamental to a coherent legal theory of scientific evidence.... [R]eliability means that a successful outcome, or a correct answer, is sufficiently probable for a given situa-tion_ In contrast to reliability, validity *1112means that which results from sound and cogent reasoning. An invalid conclusion cannot be reliable, yet valid reasoning does not necessarily lead to reliable conclusions. Reliability is the ultimate legal concern, but when it hinges on controversial contested reasoning, the validity of that reasoning must be addressed. Distinguishing between validity and reliability is important because it permits the separation of scientific questions from legal questions.... [T]he scientific question [should be viewed] as a matter of validity, with the answer depending on accepted scientific practice and the soundness and cogency of the entire pattern of reasoning leading to the expert’s conclusion. In contrast, the legal question relates to how much reliability the law requires, with the answer depending on legal standards.
Id. at 599-600.
. As Judge Learned Hand once observed, expert testimony creates the risk of a special kind of prejudice:
The trouble with all this is that it is setting the jury to decide, where doctors disagree. The whole object of the expert is to tell the jury, not facts, as we have seen, but general truths derived from his specialized experience. But how can the jury judge between two statements each founded upon an experience confessedly foreign in kind to their own? It is just because they are incompetent for such a task that the expert is necessary at all.
Hand, Historical and Practical Considerations Regarding Expert Testimony, 15 Harv.L.Rev. 40, 53 (1901).
. Rule 703 of the Federal Rules of Evidence provides that:
The facts or data in the particular case upon which an expert bases an opinion or inference may be those perceived by or made known to the expert at or before the hearing. If of a type reasonably relied upon by experts in the particular field in forming opinions or inferences upon the subject, the facts or data need not be admissible in evidence.
. See e.g., Slaughter v. Southern Talc Co., 919 F.2d 304 (5th Cir.1990); Brown v. Parker-Hannifin Corp., 919 F.2d 308 (5th Cir.1990); Edmonds v. Illinois Cent. Gulf Ry. Co., 910 F.2d 1284 (5th Cir.1990); Brock v. Merrell Dow Pharmaceuticals, 874 F.2d 307 (5th Cir.1989), cert. denied, — U.S. —, 110 S.Ct. 1511, 108 L.Ed.2d 646 (1990); Washington v. Armstrong World Indus. Inc., 839 F.2d 1121 (5th Cir.1988); Viterbo v. Dow Chemical Co., 826 F.2d 420 (5th Cir.1987); Osburn v. Anchor Laboratories, Inc., 825 F.2d 908 (5th Cir.1987); Thompson v. Southern Pac. Trans. Co., 809 F.2d 1167 (5th Cir.1987); In re Air Crash Disaster at New Orleans, Louisiana, 795 F.2d 1230 (5th Cir.1986); Soden v. Freightliner Corp., 714 F.2d 498 (5th Cir.1983). But see Peteet v. Dow Chemical Co., 868 F.2d 1428 (5th Cir.1989); Dixon v. International Harvester Co., 754 F.2d 573 (5th Cir.1985).
. Further, even if experts may generally rely on a certain source or quantum of data in the absence of contrary or more complete information, this may not suffice to admit an opinion critically resting on the former, when significantly different and more accurate or complete information is established beyond reasonable dispute in the record.
. See Black, A Unified Theory of Scientific Evidence, 56 Fordham L.Rev. 595 (1988).
. We have applied Frye in a civil, as well as a criminal, case. See Barrel of Fun, 739 F.2d 1028.
. The Christophersen panel concluded that the district court, as well as the defendants’ experts, by overemphasizing Dr. Miller's reliance on the fact that small-cell carcinoma in the colon looks similar to small-cell carcinoma in the lung, misunderstood and oversimplified Dr. Miller’s theory of causation. 902 F.2d at 366. The panel then stated that, "Dr. Miller’s analysis was based on the nature of the biochemical reaction that results in the development of small cell carcinoma.” Id. We are persuaded, however, that under the standard of review of manifest error the defense experts’ testimony and the district court’s finding were sufficiently broad to be understood as rejecting Dr. Miller’s methodology.