NSF Merit Review Panel

An NSF merit review panel is the group of external subject-matter experts the National Science Foundation convenes to evaluate proposals against its two National Science Board (NSB)-approved merit review criteria — Intellectual Merit and Broader Impacts — and to produce individual ratings, written reviews, and (for panel review) a group discussion summary that an NSF program officer weighs when recommending a funding decision. NSF review can be conducted by a panel, by mail (ad hoc) reviewers working individually, or by a combination of both, all governed by Chapter III of NSF’s Proposal & Award Policies & Procedures Guide (PAPPG).

The two criteria every reviewer applies

Since 1997, NSF has evaluated every proposal it accepts for review against the same two criteria, applied together rather than as alternatives:

• Intellectual Merit — the potential for the proposed activity to advance knowledge and understanding within its field or across different fields.
• Broader Impacts — the potential to benefit society and contribute to specific, desired societal outcomes, such as improving STEM education, broadening participation in STEM, building academia–industry partnerships, or strengthening research and education infrastructure.

NSF revised how it expects Broader Impacts to be evidenced starting in 2024, with rolling implementation through 2025–2026, weighing plans grounded in the broader-impacts research literature with measurable outcomes over generic, plausible-sounding activities — see CASRAI’s news coverage, What “broader impacts” means under the new NSF policy, and the orientation guide NSF Grants: An Overview for how merit review fits into the wider proposal lifecycle.

Panel review vs. ad hoc review

A program officer chooses the review mechanism for a given competition, and NSF’s own guidance describes three patterns:

• Panel review: a group of external reviewers (commonly assembled for a specific competition or standing program) meets — in person or, increasingly, virtually — to discuss proposals collectively after submitting individual written reviews and ratings, then rank or group proposals for the program officer.
• Ad hoc (mail) review: individual experts review and rate a proposal independently, without a group discussion, typically used to supplement panel expertise or when a program doesn’t convene a standing panel.
• Combination review: ad hoc written reviews feed into a panel’s discussion, giving the panel additional specialist input beyond its own members’ expertise.

Program officers typically identify at least three reviewers per proposal. Reviewers submit individual written reviews addressing both criteria before any panel meets, so a panel’s discussion builds on independent assessments rather than starting from a blank page.

The rating scale

Reviewers rate each proposal Excellent, Very Good, Good, Fair, or Poor (sometimes mapped to 5–1 for quantitative summaries), with definitions running roughly: Excellent (outstanding in all respects, highest priority for support), Very Good (high quality in nearly all respects, should be supported if possible), Good (a quality proposal worthy of support), Fair (lacking in one or more critical aspects, key issues need addressing), and Poor (serious deficiencies). A rating is not itself a funding decision — it is input the program officer weighs alongside program-level considerations such as portfolio balance, potential for transformative results, and available funds.

From review to award recommendation

After individual reviews (and, for panels, a group discussion) are complete, the program officer synthesizes reviewer input into a recommendation, which then requires concurrence from the relevant Division Director before an award can be processed. A favorable set of reviews is necessary but not sufficient: NSF’s overall funding (award) rate across all directorates typically runs in the mid-to-high twenties as a percentage of proposals received, and the program officer’s programmatic judgment — not a mechanical ranking of ratings — determines which well-reviewed proposals actually get funded within a limited budget.

Reviewer conflicts of interest

NSF program officers are responsible for identifying potentially disqualifying conflicts of interest before a reviewer is assigned a proposal. Under NSF’s conflict-of-interest policy, a reviewer generally cannot review a proposal if the reviewer (or a spouse, minor child, or business partner) has a financial interest in the outcome, or where the reviewer has a close personal or professional relationship with an investigator — a current or recent collaborator, a former PhD advisor or advisee, or a close relative, among other disqualifying relationships listed in the PAPPG. Panelists disclose potential conflicts on NSF’s Conflict-of-Interests and Confidentiality Statement for NSF Panelists (NSF Form 1230P) and recuse themselves from discussion and voting on any proposal where a conflict applies.

NSF merit review vs. NIH study section: not the same mechanism

NSF’s merit review panel is frequently, and incorrectly, treated as a direct equivalent of NIH’s Study Section (Scientific Review Group) process, and the two are structurally different in ways that matter for anyone administering or comparing awards across both agencies:

• Criteria: NSF uses exactly two NSB-mandated criteria, Intellectual Merit and Broader Impacts, applied to every proposal agency-wide. NIH study sections score scientific and technical merit against several weighted core review criteria (significance, investigators, innovation, approach, environment) that vary by activity code, producing a numerical priority score rather than a five-point letter-grade-style rating.
• Administering body: NIH study sections are convened and staffed centrally, primarily by NIH’s Center for Scientific Review (CSR); NSF merit review panels are convened directly by the cognizant program officer within each NSF directorate and division, with no single foundation-wide review-coordination office equivalent to CSR.
• Structure: NIH distinguishes chartered (standing, multi-year-member) study sections from temporary Special Emphasis Panels. NSF’s panel/ad hoc/combination distinction is a different axis entirely — it describes how a single competition is reviewed, not a standing-vs-temporary roster distinction.
• Second-level review: an NIH application that scores well at study section still faces a second-level programmatic review by the funding Institute or Center’s advisory council. NSF’s equivalent second check is the Division Director’s concurrence on the program officer’s recommendation — a narrower, faster step than NIH’s council-meeting cadence.

The two processes are also not the same thing as an institutional Scientific Review Committee (SRC), which reviews an already-funded clinical trial protocol’s scientific merit and feasibility at the institution running the study — a different object of review (a protocol, not a funding proposal) at a different point in the research lifecycle, most common at NCI-designated cancer centers.

Where merit review fits in the broader proposal lifecycle

Merit review is one step within NSF’s pre-award process, which is itself shaped by the government-wide Uniform Guidance (2 CFR 200) — see CASRAI’s guide to 2 CFR 200 Subpart C: Pre-Federal Award Requirements for how federal agencies generally structure funding-opportunity publication, application review, and pre-award risk assessment. For how NSF’s specific review approach compares with NIH and the European Research Council on one live topic — disclosure of generative-AI use in proposal preparation and review — see CASRAI’s comparison NIH vs NSF vs ERC: AI Policies for Grant Writing.

References

• NSF, “How We Make Funding Decisions” — overview of the two merit review criteria and the review process.
• NSF, PAPPG (NSF 24-1), Chapter III: NSF Proposal Processing and Review — governing rules for panel, ad hoc, and combination review, and the rating scale.
• NSF, “Conflicts of Interest” policy page — disqualifying financial and personal-relationship conflicts for reviewers, and NSF Form 1230P.