The Gatekeeper Reliability Hearing (Three-Prong Kelly Test)

What Is a Kelly Gatekeeping Hearing, and How Do You Get One?

A Kelly hearing is a pretrial proceeding in which the judge, acting as gatekeeper, decides whether scientific or forensic expert testimony is reliable enough to reach the jury. In Texas, the defense triggers it by objecting to the State’s expert and asking the court for a reliability determination under Rule of Evidence 702.

Texas Rule of Evidence 702 permits a qualified expert to testify only when scientific, technical, or other specialized knowledge will help the jury. The threshold question is whether the evidence is helpful, and to be helpful it must be reliable. As the Court of Criminal Appeals put it in Somers v. State, the trial court “must act as a gatekeeper to ensure that unreliable evidence does not reach the trier of fact.”

The gatekeeping standard is not reserved for exotic or brand-new techniques. In Hartman v. State, the court held that the Kelly reliability standard applies to all scientific evidence offered under Rule 702, not merely “novel” science. Whether the State offers DNA typing, a breath-alcohol extrapolation, or a firearms toolmark comparison, the same reliability screen applies.

The defense ordinarily triggers the hearing by filing a written motion to exclude the expert and objecting when the witness is offered, asking the court to hold a gatekeeping hearing before the testimony is heard. The trial court has discretion over “whether, when, and how” it conducts that inquiry, but it may not skip the reliability determination once a proper objection is made. Filing early, well before trial, gives the court time to hear competing experts and study the literature.

Not every witness triggers a full Kelly hearing. The gatekeeping inquiry is aimed at scientific, technical, and specialized-knowledge testimony — the laboratory result, the crash reconstruction, the medical-causation opinion. An officer merely describing what he saw is not offering expert science. The defense identifies which of the State’s witnesses are truly experts, then targets the reliability of each opinion the State intends to elicit, because a favorable ruling on one discipline does not resolve the others.

The Three Prongs: Theory, Technique, and Application

Under Kelly v. State, scientific evidence is reliable only if it satisfies three criteria: the underlying scientific theory is valid, the technique applying that theory is valid, and the technique was properly applied on the occasion in question. All three must be shown, and a failure on any one prong defeats admission.

Kelly discarded the old Frye “general acceptance” test in favor of a Rule 702 reliability analysis. The court reasoned there is no textual basis in Rule 702 for a special standard, and that evidence can be reliable even before it is generally accepted. What replaced Frye was the three-part inquiry that Texas courts still apply today, restated as recently as Rhomer v. State.

The prongs are cumulative. A theory can be impeccable and the technique valid, yet the evidence still fails if the analyst botched the protocol in the specific case — the third prong. Conversely, a flawless application cannot rescue a theory the scientific community rejects. Because the burden runs prong by prong, the defense can prevail by breaking any single link in the chain rather than dismantling the entire field.

Framing the challenge around the weakest prong is a matter of strategy. For a mature discipline like DNA typing, the first two prongs are usually conceded or judicially noticed, so the defense concentrates its fire on the third — contamination, mixture interpretation, or a deviation from the laboratory’s own protocol in this case. For a newer or more subjective method, the theory and technique prongs themselves are contestable, and counsel builds a record showing the field lacks the validation Rule 702 demands.

The Seven Kelly Reliability Factors

To help judges decide the first two prongs, Kelly listed seven non-exclusive factors, ranging from acceptance in the scientific community to the technique’s potential rate of error. No single factor controls. The judge weighs them flexibly, and some factors may not fit a given field at all.

The seven factors a Texas court may consider are:

1. the extent to which the theory and technique are accepted as valid by the relevant scientific community, if one can be identified;
2. the qualifications of the experts testifying;
3. the existence of literature supporting or rejecting the theory and technique;
4. the potential rate of error of the technique;
5. the availability of other experts to test and evaluate the technique;
6. the clarity with which the theory and technique can be explained to the court; and
7. the experience and skill of the person who applied the technique on the occasion in question.

The error rate and community acceptance usually carry the most weight, because a method with a high or unknown error rate is, by definition, unreliable. These factors track the federal gatekeeping framework of Daubert v. Merrell Dow Pharmaceuticals, which asks whether a theory can be and has been tested, whether it has been peer reviewed and published, its known rate of error, and its general acceptance. Texas courts treat Daubert as persuasive but apply the Kelly formulation.

No single factor is dispositive, and a technique need not satisfy all seven to be admitted. The judge weighs them together, and their relative importance shifts with the discipline. For a measurement-based method, error rate and controlling standards dominate; for a method that turns on an examiner’s interpretation, the availability of validating literature and independent testing carries more weight. Because the list is expressly non-exclusive, the defense may also raise field-specific reliability concerns that Kelly never enumerated.

Who Bears the Burden, and Why the Jury Steps Out

The proponent of the evidence — in a criminal case, almost always the State — bears the burden of proving reliability by clear and convincing evidence. The hearing takes place outside the jury’s presence, so jurors never hear the disputed science unless the judge first finds it reliable and admissible.

Kelly is explicit on both points: the proponent must persuade the trial court, “by clear and convincing evidence, that the evidence is reliable and therefore relevant,” and all three criteria “must be proven to the trial court, outside the presence of the jury, before the evidence may be admitted.” That procedure flows from Rules 104(a) and (c), which assign preliminary admissibility questions to the judge. Somers v. State reaffirmed that the State carries this clear-and-convincing burden.

The defense has a distinct procedural tool at the hearing. Rule 705(b) gives the opponent the right, on request, to voir dire the expert about the underlying facts and data outside the jury’s presence before the expert states an opinion in front of the jury. That examination is where counsel pins the analyst down on protocol deviations, missing controls, and the limits of the method — the raw material for both an exclusion argument and, if the evidence comes in, a devastating cross-examination. Even when the judge admits the testimony, the defense may still cross-examine and offer a competing expert before the jury.

The stakes of the hearing reach past a simple keep-it-out or let-it-in ruling. Even where the court admits the evidence, the sworn testimony the expert gives at the hearing is locked in, and any inconsistency between that testimony and what the expert later tells the jury becomes impeachment. A thorough Rule 705(b) examination therefore pays dividends either way — it can win exclusion, and if it does not, it arms the cross-examination the jury will hear.

Which Test Applies: Hard Science Versus Soft Science?

The first thing a reliability hearing decides is which test governs. The rigorous Kelly factors apply to the hard sciences; fields built on experience and training are measured by the more flexible three-part inquiry from Nenno v. State. Choosing the wrong track is itself a litigable issue that can affect the outcome on appeal.

In Rhomer v. State, the Court of Criminal Appeals held that the Nenno test, not the Kelly factors, governed an accident-reconstruction opinion because the expert relied on his training and experience evaluating physical evidence at crash scenes rather than on a hard scientific calculation such as pre-impact speed. The point is that the same courtroom question — is this reliable? — is measured differently depending on the nature of the discipline.

The federal courts reach the same structural result. In Kumho Tire Co. v. Carmichael, the Supreme Court held that the gatekeeping obligation applies to all expert testimony based on technical or other specialized knowledge, not just scientific testimony, though the specific reliability factors flex with the field. Texas practitioners argue the hard-versus-soft classification early, because it dictates what the State must prove. A companion page in this guide covers the soft-science Nenno and Medrano standard in detail.

Getting the classification right is not academic. If a court applies the relaxed Nenno inquiry to what is really hard science, it may admit a method without demanding the error-rate and validation proof Kelly requires; if it applies rigid Kelly factors to an experience-based field, it may exclude reliable testimony for lacking a rate of error the field cannot produce. Counsel should brief the classification question directly and, where the science is mixed, explain which parts fall on each side of the line.

Judicial Notice, Preservation, and Appellate Review

Once a technique’s theory and method have been validated in enough contested hearings, courts may take judicial notice of the first two prongs, sparing the proponent from re-proving them. But the third prong — correct application in this case — is always open, and a trial court’s reliability ruling is reviewed only for abuse of discretion.

Somers v. State explained that after a scientific theory or technique is widely accepted through adversarial gatekeeping hearings in enough trial courts, later courts may take judicial notice of its validity. That is why, for example, the reliability of the theory and technique behind certain established tests is treated as settled. But judicial notice reaches only the first two prongs; whether the technique was correctly applied to the defendant’s sample must still be litigated case by case.

That reality shapes strategy. Even against a well-established method, the strongest challenge is usually the third prong — the analyst’s deviations, contamination, calibration gaps, or documentation failures in the specific case. To preserve the issue for appeal, counsel must obtain a clear ruling on the record. Reviewing courts apply an abuse-of-discretion standard and uphold a ruling that falls within the zone of reasonable disagreement, evaluating it on what was before the trial judge when the ruling was made. A well-built record at the hearing therefore does double duty: it can exclude the evidence, and it frames any later appeal.

Preservation is unforgiving. To complain on appeal that unreliable evidence came in, the defense must have lodged a timely, specific objection and obtained a ruling; a general “it is not reliable” objection may not preserve a targeted third-prong complaint. And because review is deferential when the State prevailed below, reversal is difficult. That asymmetry is why the contest over forensic science is usually won or lost in the trial court rather than on appeal.

Discovery, Defense Experts, and the Hearing Itself

A reliability hearing is only as strong as its preparation. The defense builds one by obtaining the analyst’s underlying records, retaining an expert where the discipline warrants it, and using the hearing to expose gaps in validation and application before the jury ever hears the disputed science.

The raw material for a Kelly challenge lives in documents the State does not volunteer: the analyst’s bench notes, the laboratory’s validation studies and standard operating procedures, calibration and maintenance logs, proficiency-test results, and any corrective-action reports. Discovery under Article 39.14 and targeted requests give the defense access to these materials, which frequently reveal the protocol deviations that defeat the third prong or the thin validation that undermines the first two.

Where the discipline is complex, a defense expert can be decisive. The expert translates the literature for the court, identifies the recognized standards the State’s witness ignored, and offers a competing account of error rate and reliability. Even when the court ultimately admits the State’s evidence, a credible opposing expert reshapes how the jury weighs it.

The hearing itself is the arena. Counsel uses Rule 705(b) to take the expert on voir dire outside the jury, walks the witness through each prong, and secures concessions on the record. The written motion should lay out the three-prong framework, identify precisely which prong fails and why, and ask for a specific ruling. Done well, the hearing either keeps the evidence from the jury or produces a transcript that guides an effective cross-examination when the jury returns.

Where This Fits

This guide is one of four situations that arise under the Motion to Exclude Expert / Kelly-Daubert Challenge. Start with the parent motion for the overall framework, or move to a related fact pattern:

What the Case Law Says

These decisions—verified against primary sources—control how this issue is litigated. Every case still turns on its own facts.

• Kelly v. State, 824 S.W.2d 568 (Tex. Crim. App. 1992) — Scientific evidence is reliable only if the theory and technique are valid and properly applied, proven by clear and convincing evidence outside the jury.
• Hartman v. State, 946 S.W.2d 60 (Tex. Crim. App. 1997) — The Kelly reliability standard applies to all scientific evidence under Rule 702, not just novel science; judges are gatekeepers who weed out junk science.
• Somers v. State, 368 S.W.3d 528 (Tex. Crim. App. 2012) — The proponent must persuade the court by clear and convincing evidence that scientific evidence is reliable; courts may judicially notice validated theory and technique.
• Rhomer v. State, 569 S.W.3d 664 (Tex. Crim. App. 2019) — Courts first decide which reliability test applies; the flexible Nenno test governed experience-based accident reconstruction rather than the hard-science Kelly factors.
• Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993) — Trial judges must ensure scientific testimony is reliable, weighing testability, peer review, error rate, and general acceptance under a flexible inquiry.
• Kumho Tire Co. v. Carmichael, 526 U.S. 137 (1999) — The gatekeeping duty extends to all expert testimony based on technical or specialized knowledge, though the specific reliability factors flex with the field.

General summaries of published opinions for information only — not predictions about any specific case.