What Is a Clinical Trial? NIH's Definition

NIH does not define a “clinical trial” by phase, by whether a drug or device is involved, or by whether the intervention is already approved. It uses a four-question operational test, and answering “yes” to all four makes a study an NIH-defined clinical trial — regardless of how small, early-stage, or non-regulatory the study is. This is the single most consequential classification decision in an NIH application: it determines which funding opportunities are open to a study, whether it must register and report on ClinicalTrials.gov, whether staff need Good Clinical Practice (GCP) training, and which application form is required. It is also a narrower, more specific test than researchers coming from an ICMJE or FDA background often assume — and it changed materially for 2026.

NIH’s definition of a clinical trial

NIH defines a clinical trial as:

“A research study in which one or more human subjects are prospectively assigned to one or more interventions (which may include placebo or other control) to evaluate the effects of those interventions on health-related biomedical or behavioral outcomes.”

This definition sits in NIH’s Grants Policy Statement and grants.nih.gov guidance, and it is deliberately broad on some dimensions and narrow on others. It is phase-agnostic (a Phase 1 dose-finding study can qualify), intervention-agnostic (drugs, devices, behavioral interventions, educational strategies, and dietary interventions can all qualify), and it applies even when the “intervention” is a diagnostic or a public health strategy rather than a therapeutic product. What it requires, non-negotiably, is prospective assignment to an intervention and a plan to evaluate an effect on a health-related outcome. Purely observational or naturalistic data collection does not meet it, no matter how clinically relevant the topic is.

The four-question test

NIH’s own decision framework reduces the definition to four yes/no questions. A study is an NIH-defined clinical trial only if the answer to all four is yes:

Does the study involve human participants?
Are the participants prospectively assigned to an intervention?
Is the study designed to evaluate the effect of the intervention on the participants?
Is the effect being evaluated a health-related biomedical or behavioral outcome?

A single “no” anywhere in that chain means the study does not meet NIH’s definition, even if it is still human subjects research subject to IRB review and other protections. NIH publishes a formal decision tool and a compiled set of case-study scenarios (mechanistic studies, secondary data analyses, training/education interventions, and more) that work through this test against realistic proposals — useful for the genuinely ambiguous cases, which are common at the boundary of basic and applied research.

What typically answers “yes” to all four

A randomized trial testing a drug, device, or biologic against placebo or standard of care.
A Phase 1 study assessing only pharmacokinetics, safety, or maximum tolerated dose of an investigational product — NIH’s guidance is explicit that this still counts, even though it involves no efficacy claim.
A behavioral intervention trial (e.g., a coaching, counseling, or digital-therapeutic protocol) that prospectively assigns participants to conditions and measures a health-related behavioral outcome.
A mechanistic clinical trial that assigns an intervention to understand a biological process and measures a health-related biomedical or behavioral outcome as part of that assessment.

What typically answers “no” to at least one question

A purely observational cohort or case-control study with no assigned intervention (fails question 2).
A secondary analysis of existing data or specimens with no new prospective assignment (fails question 2).
A survey, chart review, or registry study describing practice patterns without assigning an intervention (fails question 2).
A study assigning an intervention purely to understand a mechanism, with no health-related biomedical or behavioral outcome evaluated — as of the 2026 policy change described below, this category (Basic Experimental Studies Involving Humans) is treated separately from NIH-defined clinical trials even though it used to meet all four questions.

Why the classification matters: what it triggers

Being an NIH-defined clinical trial is not a label with no consequences — it activates a specific set of NIH policy and application requirements that do not apply to other human subjects research:

Funding opportunity eligibility

NIH funding opportunity announcements (FOAs) are coded Clinical Trial Required, Clinical Trial Optional, or Clinical Trial Not Allowed. Submitting an NIH-defined clinical trial to a “Not Allowed” mechanism, or a non-trial study to a “Required” one, results in the application being withdrawn from review without a scientific evaluation — this is an administrative gate, checked before peer review even begins.

ClinicalTrials.gov registration and results reporting

Under the NIH Policy on the Dissemination of NIH-Funded Clinical Trial Information (effective 18 January 2017), every NIH-funded study that meets NIH’s own clinical trial definition must be registered and have results reported on ClinicalTrials.gov — a broader population than the FDA Amendments Act’s narrower “applicable clinical trial” category, which excludes Phase 1 studies entirely. See CASRAI’s clinical trial registration and reporting compliance guide for how the NIH, FDAAA, and ICMJE registration obligations differ and where they overlap.

Good Clinical Practice (GCP) training

Since 1 January 2017, NIH policy requires that all NIH-funded investigators and staff responsible for the conduct, management, or oversight of an NIH-defined clinical trial complete Good Clinical Practice (GCP) training, refreshed at least every three years, regardless of the trial’s phase or intervention type.

Single IRB review for multi-site studies

Separately, NIH’s single-IRB (sIRB) policy (effective 25 January 2018) requires domestic sites of NIH-funded multi-site, non-exempt human subjects research using the same protocol to rely on one IRB of record rather than each site obtaining independent local review, subject to specific exceptions. This policy applies to qualifying multi-site human subjects research generally, not exclusively to studies that meet the four-question clinical trial test — but NIH-defined clinical trials are its central, paradigm case. CASRAI’s IRB/REC approval process guide covers what does and doesn’t require review.

The PHS Human Subjects and Clinical Trials Information Form

Since applications with due dates on or after 25 January 2018 (the FORMS-E application package and later), NIH applications proposing human subjects research use a single, consolidated PHS Human Subjects and Clinical Trials Information form. For studies that meet NIH’s clinical trial definition, this form requires a full protocol synopsis, study design and outcome fields, and other detail not required of non-trial human subjects research.

How this differs from the FDA’s and ICMJE’s definitions

Researchers frequently assume “clinical trial” means the same thing everywhere it appears in a research-administration context. It doesn’t. Three distinct bodies apply three different definitions, and a study can be a clinical trial under one and not another:

ICMJE’s definition is broader still than NIH’s on paper — “any research project that prospectively assigns people or a group of people to an intervention… to study the relationship between a health-related intervention and a health outcome” — and explicitly includes Phase 1 trials, matching NIH’s inclusion of Phase 1. ICMJE requires prospective registration as a condition of publication in its member journals, independent of any funder.
FDAAA 801’s “applicable clinical trial” (ACT) is narrower than NIH’s definition: controlled trials of drugs or biologics other than Phase 1, plus certain device trials. Phase 1 studies are explicitly excluded from ACT status, even though they are explicitly included under both the NIH and ICMJE definitions. This is a genuine, frequently-confused point of divergence — a Phase 1 NIH-funded drug study can be an NIH-defined clinical trial (register/report under NIH policy) while not being an FDAAA “applicable clinical trial” at all.
NIH’s definition, described on this page, governs which NIH grant, cooperative agreement, or contract requirements apply — independent of, though frequently overlapping with, both of the above.

See CASRAI’s dictionary entries on clinical trial registration and prospective clinical trial registration, and the ICMJE dictionary entry, for the registration-specific mechanics that follow from each definition.

Not the same as Medicare’s “qualifying clinical trial” definition

A separate, unrelated definition shows up in clinical-research billing: the Centers for Medicare & Medicaid Services’ National Coverage Determination for Routine Costs in Clinical Trials (NCD 310.1) defines a “qualifying clinical trial” for the purpose of determining whether Medicare will cover the routine patient-care costs of a beneficiary enrolled in a trial. NCD 310.1’s qualifying criteria (subject evaluates a Medicare benefit category, is therapeutic in intent, and meets one of several deeming criteria including NIH funding) are a billing-compliance test, not a grants-administration test, and answering NIH’s four-question test does not automatically answer CMS’s. Research administrators working across both grants management and clinical trial billing should treat these as two separate determinations, made for two separate purposes, even though the same study is often subject to both.

2026 update: Basic Experimental Studies Involving Humans (BESH) are no longer classified as clinical trials

This is the most significant recent change to how the NIH definition is applied in practice. NIH previously required investigators to classify any study that met both (a) NIH’s definition of basic research (systematic study toward greater understanding of fundamental phenomena, without a specific application in mind) and (b) all four clinical-trial questions above as a Basic Experimental Study Involving Humans (BESH) — and BESH studies were treated as a subtype of NIH-defined clinical trial, subject to the same registration, reporting, and FOA-type restrictions as therapeutic trials. This produced years of documented friction: purely mechanistic studies (for example, a study assigning a cognitive task to measure a neural or behavioral response, with no intent to evaluate or change clinical practice) were required to register on ClinicalTrials.gov and comply with clinical-trial-specific application requirements designed for therapeutic research.

NIH announced a change to this via NOT-OD-26-032 (29 January 2026), with implementation detail in NOT-OD-26-067 (6 April 2026): effective for applications with due dates on or after 25 May 2026, BESH studies are no longer classified as NIH-defined clinical trials. Practically, this means:

BESH studies no longer trigger ClinicalTrials.gov registration or results reporting.
BESH studies no longer need to be submitted to “Clinical Trial Required” FOAs, and can be submitted under “Clinical Trial Not Allowed” mechanisms provided the application does not also propose a separate NIH-defined clinical trial.
Human subjects protections (IRB review, informed consent) and other generally applicable NIH policies — including the NIH Data Management and Sharing Policy — continue to apply in full; this is a reclassification of clinical-trial-specific administrative requirements, not an exemption from human subjects oversight.
Already-funded BESH projects continue under the terms and conditions in place at the time of award; institutions should confirm with grants management staff whether an in-progress study now qualifies as BESH under the revised criteria before assuming the change applies retroactively.

Because this changed only months before this page was published, verify current guidance directly against NIH’s BESH policy page and the two notices above before relying on it for a specific submission decision — implementation FAQs and institute-specific guidance are still being issued.

How to determine whether your study meets the definition

For a specific, in-development protocol, don’t rely on analogy to a similar-sounding study. NIH provides two free, authoritative tools:

A decision tool that walks through the four questions against your specific design.
A compiled set of case studies covering common edge cases — mechanistic studies, studies using existing data or specimens, and studies involving training or educational interventions among them.

Where the classification is genuinely ambiguous, institutional research administration or sponsored-programs offices generally route the determination through the same office that manages human subjects research exemption determinations, since both draw on the same protocol details.

Frequently asked questions

Is a Phase 1 trial an NIH-defined clinical trial?

Yes. NIH’s guidance is explicit that a study designed only to assess pharmacokinetics, safety, or maximum tolerated dose of an investigational product still meets the definition if it prospectively assigns participants to an intervention and evaluates a health-related outcome. This is a key difference from FDAAA 801, whose narrower “applicable clinical trial” category excludes Phase 1 studies.

Does NIH’s definition include behavioral and social-science interventions?

Yes. The definition is intervention-agnostic: behavioral, educational, and public-health interventions can meet the four-question test in the same way a drug or device trial can, provided there is prospective assignment and a health-related biomedical or behavioral outcome being evaluated.

Is an observational study ever an NIH-defined clinical trial?

No. Purely observational designs — cohort studies, case-control studies, registries, and secondary data analyses with no prospective assignment to an intervention — fail question 2 of NIH’s test and are not clinical trials under this definition, whatever else they may be subject to (IRB review, data sharing requirements, and so on).

Are Basic Experimental Studies Involving Humans (BESH) still clinical trials?

Not as of applications due on or after 25 May 2026. NIH reclassified BESH studies out of the clinical-trial category via NOT-OD-26-032 and NOT-OD-26-067; prior to that change, a BESH study was treated as a subtype of NIH-defined clinical trial.

Does meeting NIH’s definition automatically satisfy CMS’s Medicare coverage requirements, or vice versa?

No. NIH’s four-question test and CMS’s NCD 310.1 “qualifying clinical trial” test serve different purposes (grants administration vs. Medicare billing for routine patient-care costs) and use different criteria. A study can meet one without automatically meeting the other; both determinations need to be made separately.

Where can I get an authoritative determination for my own study?

Use NIH’s own definition page and decision tool, and route genuinely ambiguous cases through your institution’s sponsored-programs or research-compliance office, which typically coordinates with the IRB on the same protocol facts.