Exclusion is the exception, and usually not the point
The honest starting proposition is one most writing on this subject omits: a substantial majority of testifying experts are never challenged at all, and among those who are, exclusion in whole is uncommon. Exclusion in part — some opinions struck, others narrowed, the rest admitted — is the ordinary shape of a win.
That is not an argument against challenging. It is an argument for clarity about what a challenge is for. The realistic objectives, in descending order of how often they are achieved: narrowing the opinion so the expert cannot roam at trial; obtaining a limiting instruction; forcing the proponent to commit in writing to what the methodology actually is; generating the cross-examination; and, least often, exclusion.
Every one of those except the last survives a motion that is denied. A party that files nothing and cross-examines at trial gets none of them.
The line the whole exercise runs along
Every expert challenge is an argument about which side of one line a defect falls on. Defects going to admissibility are for the judge under FRE 104(a), decided by a preponderance, and can keep an opinion from the jury entirely. Defects going to weight are for the jury, and FRE 104(e) preserves the right to present them: the rule “does not limit a party's right to introduce before the jury evidence that is relevant to the weight or credibility of other evidence.”
The sentence that has carried more opposition briefs than any other comes from Daubert, 509 U.S. at 596: “Vigorous cross-examination, presentation of contrary evidence, and careful instruction on the burden of proof are the traditional and appropriate means of attacking shaky but admissible evidence.” The load-bearing words are “shaky but admissible.” It describes what to do with testimony that has already passed the gate, and says nothing about which testimony passes.
What the 2023 amendment did to that line
Effective 1 December 2023, FRE 702 requires the proponent to demonstrate “that it is more likely than not” that each of the rule's four requirements is satisfied, and subsection (d) now asks whether the opinion “reflects a reliable application” of the principles and methods to the facts of the case. The Advisory Committee explained why: “many courts have held that the critical questions of the sufficiency of an expert's basis, and the application of the expert's methodology, are questions of weight and not admissibility. These rulings are an incorrect application of Rules 702 and 104(a).”
The amendment did not abolish the weight category. It relocated the boundary. Ordinarily still weight: that the expert's credentials are less impressive than the opponent's; that a well-qualified opposing expert reached a different conclusion; that a contrary study exists which the expert acknowledges and addresses; a judgment call made within an accepted method; compensation and retention history; a minor computational error that does not drive the result.
Ordinarily admissibility after 2023: that the expert's field does not cover the opinion offered; that the stated methodology was not actually applied to these facts, under 702(d); that the facts or data are insufficient to support the opinion at all, under 702(b); that data were used selectively in a way that drives the result; that the method exists only for this litigation; and that the conclusion is connected to the data only by the expert's say-so.
Where the line sits now, and who disagrees
The circuits are visibly re-sorting decisions across that boundary, and the process is not finished. The Sixth Circuit in In re Onglyza (Saxagliptin) and Kombiglyze XR Products Liability Litigation, 93 F.4th 339 (6th Cir. 2024), affirmed exclusion of an MDL general-causation expert whose opinion rested heavily on a single finding within one trial while discounting contrary evidence — treating selective reliance on data as an admissibility problem.
Running the other way is Sommerville v. Union Carbide Corp., 149 F.4th 408 (4th Cir. 2025), which reversed an exclusion under 702(b) and held that questions about an opinion's factual underpinnings affect weight and credibility rather than admissibility, over a dissent from the Chief Judge. It is the decision to cite for the proposition that the re-sorting is not complete.
The Supreme Court was asked to settle it and declined. The defendants petitioned for certiorari as Union Carbide Corp. v. Sommerville, No. 25-919, presenting the question whether challenges to an expert's factual basis always go to weight or instead to admissibility where the court first finds the evidentiary support insufficient. The petition was docketed 4 February 2026 and denied on 26 May 2026, Justice Alito taking no part. A cert denial is not a merits ruling and approves nothing — but the post-2023 divergence remains unresolved nationally, with the Fourth Circuit the visible outlier.
The most-cited survey of the 2024–25 decisions is Lee Mickus, “Amended Rule 702 in 2025: Circuit Courts Embrace the Changed Standard,” published by the Washington Legal Foundation on 7 October 2025 — a body that files amicus briefs predominantly on the defense side of products liability, so its citations are checkable but its framing is advocacy. It identifies the Eighth Circuit's Sprafka v. Medical Device Business Services, 139 F.4th 656 (8th Cir. 2025), as abandoning that circuit's rule that an opinion's factual basis goes to credibility rather than admissibility.
What the empirical work supports, and what it does not
Published exclusion rates range from roughly a quarter to roughly two-thirds, and the range is an artifact of what is being counted. What is the denominator — challenges filed, or written opinions ruling on them? Most studies code published opinions, which oversamples contested and appealed rulings and undersamples the routine denial from the bench. Does “excluded” include partial exclusion — one opinion struck out of six — counted identically to wholesale exclusion? That coding choice alone moves the reported rate by tens of points. And who is counting, and what do they sell?
The named studies are these. The RAND Institute for Civil Justice study by Lloyd Dixon and Brian Gill (RAND MR-1439, 2002) coded 399 federal district court opinions addressing the reliability of expert evidence from 1980 to 1999, and reported that the proportion found unreliable rose from about 53% in the two years before Daubert to about 70% between mid-1996 and mid-1997, then leveled off; the denominator is written opinions, not all challenges. The Federal Judicial Center study by Krafka, Dunn, Johnson, Cecil and Miletich (2002) surveyed federal district judges in 1998 against a comparable 1991 survey and found them reporting more pretrial scrutiny and less admission than before Daubert — self-report data whose value is the comparison, not any absolute rate.
PwC's annual Daubert Challenges to Financial Experts series codes written decisions on challenges to accountants, economists and appraisers found in a commercial legal database; for 2021 it reported exclusion in whole or part of appraisers in 38%, accountants in 32% and economists in 27% of challenges. PwC's forensic practice supplies financial expert witnesses, and the available reporting does not state how those decisions were identified or how many were reviewed. Lawyers for Civil Justice reviewed 1,059 federal district court decisions from 2020 and found only 35% mentioned the preponderance standard and 13% stated that a presumption of admissibility exists — a study of how courts articulated the standard, not of exclusion rates, produced by an organization of corporate defense counsel to support the amendment that became the 2023 change.
No post-2023 empirical study of exclusion rates appears to exist. Everything published on the effect of the amendment is qualitative case-survey work, most of it from defense-aligned organizations. Anyone quoting a percentage for “how often Daubert motions succeed” after December 2023 is quoting something that has not been measured.
The order a challenge is actually built in
A challenge is not a decision made at a deadline; it is a sequence in which each stage produces the material the next stage needs.
Disclosure. FRCP 26(a)(2)(B) fixes the opinions and their stated bases in a signed report, along with qualifications, publications, compensation, and the list of cases in which the witness testified in the previous four years — the roadmap to prior testimony and prior exclusions.
Report analysis. The gap between what the report says the method is and what it shows the expert doing is where the motion will live. It is also when a rebuttal expert is retained if one is needed, the rebuttal deadline being short.
Deposition. The record for a Rule 702 motion is made here. A motion asserting that the expert failed to rule out an alternative cause, unsupported by testimony in which the expert says so, asks the court to take counsel's word for the state of a discipline.
The motion. Filed on the court's schedule, organized by the elements of Rule 702, aimed at identified opinions. The hearing, if there is one. There need not be; many courts decide on the papers.
Trial. Whatever survives is cross-examined, and FRE 104(e) preserves the right to put every rejected argument to the jury as weight.
What a failed challenge still buys
Four things, none of which are available to a party that only cross-examines at trial.
The opposition brief commits the proponent in writing to what the methodology is, which opinions rest on which data, and what the expert did and did not do. An expert whose trial testimony drifts from counsel's written description of the method has a problem that did not exist before the motion.
The deposition that supported the motion exists regardless of the ruling, and the concessions in it are usable under FRE 801(d)(1)(A) as substantive evidence rather than merely as impeachment, because the rule expressly names a deposition as a qualifying prior proceeding.
A hearing, if one is held, produces sworn testimony under focused questioning on the weakest part of the opinion.
And the judge who denies the motion has read the weaknesses — which matters at summary judgment, at the charge conference, and on any later question about the scope of the testimony.
Deciding whether to bring one
The costs are usually understated. A challenge tells the opponent exactly what the trial cross-examination will be, months in advance, while the expert can still supplement, clarify, or be quietly replaced. It consumes expert time on both sides. And a motion reading as an attack on credentials or compensation rather than on method spends credibility a later, better motion will need.
Against that, Rule 702 arguments are decided on the record as it stands at the deadline; a party holding the argument for trial makes it as an objection, in front of a jury, with nothing read in advance.
Where the analysis lands on challenging, it usually lands on retaining a rebuttal expert at the same time — the two products of a serious challenge are a motion and a competing methodology, and the motion is stronger when someone qualified in the field will state on the record what the method required and what was not done.
Primary sources for this page: Federal Rule of Evidence 702 and the 2023 Advisory Committee note; Supreme Court docket, Union Carbide Corp. v. Sommerville, No. 25-919.
Frequently Asked Questions
How often are challenged experts actually excluded?
No reliable single national figure exists. Published rates range from about a quarter to about two-thirds, and the spread comes from methodology: most studies code written opinions rather than all challenges, and most count partial exclusion identically to wholesale exclusion. What can be said honestly is that a substantial majority of testifying experts are never challenged, that among those challenged exclusion in whole is uncommon, and that exclusion in part is the usual form of a win.Is there any study of exclusion rates since the 2023 amendment?
None appears to have been published. Everything available on the amendment's effect is qualitative case-survey work, most of it produced by defense-aligned organizations with an interest in the framing. The pre-amendment studies — RAND's 2002 analysis of 399 district court opinions, the Federal Judicial Center's 1998 judge survey, PwC's annual financial-expert series and Lawyers for Civil Justice's review of 2020 decisions — measure different things and none of them measures the post-2023 landscape.Did the 2023 amendment eliminate the "goes to weight" argument?
No, it narrowed it. The Advisory Committee wrote that holdings treating sufficiency of basis and application of methodology as weight questions "are an incorrect application of Rules 702 and 104(a)," but it also preserved a genuine weight category and said that once the court finds an admissibility requirement more likely than not satisfied, further attacks go to weight. FRE 104(e) guarantees the right to make those attacks to the jury. What changed is the order of operations, not the existence of the category.Has the Supreme Court resolved where the line sits after 2023?
It declined to. In Union Carbide Corp. v. Sommerville, No. 25-919, the petition presented exactly that question — whether challenges to an expert's factual basis always go to weight or instead to admissibility. It was docketed 4 February 2026 and denied on 26 May 2026, with Justice Alito taking no part. A cert denial is not a merits ruling and should never be described as approval of the decision below; the practical effect is that the divergence among the circuits persists.Is it worth filing a motion that probably will not succeed?
Frequently, for reasons unrelated to the ruling. The opposition brief commits the proponent in writing to what the methodology is, and that becomes cross-examination material. The deposition that supports the motion is usable at trial as substantive evidence under FRE 801(d)(1)(A). A hearing produces sworn testimony on the weakest part of the opinion. And the judge has read the weaknesses before summary judgment. None of that is available to a party that only cross-examines.What are the costs of bringing a challenge?
The main one is disclosure. A challenge tells the opponent the trial cross-examination months in advance, while there is still time to supplement the report, retain a second expert, or reframe the opinion. It also consumes expert time on both sides, and a motion aimed at credentials or compensation rather than method spends credibility with the court. Timing the supporting deposition close to the dispositive-motion cut-off limits how much repair the disclosure enables.Should a challenge and a rebuttal expert be planned together?
Usually. The two products of a serious challenge are a motion and a competing methodology, and the motion is materially stronger when a qualified person in the field will state on the record what the method required and what was not done. Rebuttal disclosure deadlines are short — FRCP 26(a)(2)(D) sets a default of 30 days after the opposing disclosure absent stipulation or court order — so the decision to retain generally has to be made while the report is still being analyzed.Published