How to Respond to Grant Panel Reviews (NSF, NIH, ERC)

Your application came back unfunded, and you are holding a summary statement, a panel summary, or an ERC evaluation report. Before you open the proposal file, work out two things from that document. First, which one or two concerns actually moved the outcome. Second, whether those concerns can be answered by the same project done better, or only by a different project. Everything below is built around those two questions.

What each funder sends you

NIH: two scored factors and a shorter summary statement

For most research project grants with due dates on or after January 25, 2025, NIH reviews under a simplified framework that regroups the five regulatory criteria into three factors (NOT-OD-24-010).

• Factor 1, Importance of the Research. Significance and Innovation, scored 1 to 9.
• Factor 2, Rigor and Feasibility. Approach, scored 1 to 9.
• Factor 3, Expertise and Resources. Investigator and Environment, unscored. Reviewers mark it appropriate or identify a need for additional expertise or resources and explain the gap.

All three factors feed the overall impact score. NIH made Factor 3 a binary judgment to reduce the influence of general scientific reputation on review. The activity code list covers the R01, R21, U01 and most other research project grants. Fellowship, career development and training applications are not on it.

The summary statement has changed too. For the 2026 Advisory Councils and the January 2027 council, NIH modified review to handle a record volume of applications (NOT-OD-26-114). Most meetings discuss 30 to 35 percent of applications instead of the usual half. Applications voted into the middle third are labeled "competitive but not discussed" and are still considered for funding. The narrative summary of discussion is gone. In its place is one sentence on the degree of consensus in the vote and a list of the main score-driving points, followed by the three assigned reviewers' critiques. Only discussed applications get an overall impact score.

Those bullets are the most valuable lines you will receive. They are the committee's own account of what set the score, and your Introduction should answer each one by name.

NSF: reviews, a panel summary, and a reason

When NSF declines a proposal, the PI receives the reviews with reviewer-identifying information redacted, the panel summary if a panel was used, and a written explanation if the panel summary does not already give one (PAPPG 24-1, Chapter III). Every proposal is judged on two National Science Board criteria, Intellectual Merit and Broader Impacts, and the PAPPG says each is necessary while neither is sufficient alone.

There are no factor scores to sort by, so sort by repetition. The concern that appears in the panel summary and again in two individual reviews is the one to fix first. For the ways panels tend to misapply the two criteria, see our breakdown of NSF merit review in practice.

ERC: a letter grade and an interview

Under the ERC Work Programme 2026, Starting, Consolidator and Advanced Grants are evaluated in two steps. Step 1 assesses Part I of the scientific proposal and the PI's CV and track record. At Step 2 the full proposal is assessed and the PI presents it at an interview with the panel. The only criterion is excellence, applied to the project and the PI together. Step 2 ends in an A, which means the proposal is recommended for funding if the budget allows, or a B, which means it meets some but not all elements of the criterion and will not be funded.

Reading an annotated summary statement (illustrative)

The excerpt below is invented for this guide. It follows the shorter format NIH describes for the 2026 councils, for a discussed R01 on readmission after cardiac surgery in older adults. Every number in it is made up.

Overall impact. 38

Summary of discussion. The committee vote showed moderate consensus. Main score-driving points are listed below.

- Strong rationale. Readmission after cardiac surgery is an important problem, and the preliminary cohort data are convincing.

- Aim 2 depends on Aim 1 producing a positive result, and no alternative path is described.

- The primary outcome instrument has not been validated in older surgical patients.

- The analysis plan does not say how a null result in Aim 1 would be interpreted.

- Innovation is modest. The intervention adapts an established approach to a new setting.

Factor 1, Importance of the Research. Reviewers gave 3, 3 and 4.

Factor 2, Rigor and Feasibility. Reviewers gave 5, 6 and 5.

Factor 3, Expertise and Resources. Appropriate, all reviewers.

Reviewer 2, Factor 2. The team is well qualified and the cohort is a real asset. Enthusiasm is reduced by the dependency between the aims. If the Aim 1 effect is small, Aim 2 as written cannot be interpreted. The outcome measure was developed in community-dwelling adults, and its validity after major surgery is not established.

Reviewer 3, Factor 1. The question matters. The approach is a careful adaptation of a known intervention, and the application would be stronger if it explained what the adaptation reveals about mechanism.

Here is what to mark, in order.

1. The bullets outrank everything else. The first one is a strength, which tells you what to protect. The other four are your revision list.
2. Read the factor scores as a shape. Factor 2 trails Factor 1 by one to three points for every reviewer, and Factor 3 drew no concerns. The panel believes the question is worth asking and that this team can answer it. It doubts the design.
3. "Enthusiasm is reduced by" names a score driver. Whatever follows that phrase is what the reviewer would cite if asked why the score was not better. Here it is the aim dependency, which also appears in the bullets. Any concern that shows up in both places goes first.
4. Separate causes from symptoms. The aim dependency and the missing null-result plan read like two concerns. Both come from one decision, which was to write Aim 2 as if Aim 1 could only succeed.
5. Notice what is absent. Nobody questioned the cohort, the team or the environment. Do not spend revision pages there.
6. Mild praise can hide a ceiling. "The question matters" next to "careful adaptation" means Innovation is capping Factor 1. NIH's criteria say a project that does not apply novel concepts may still be of critical importance to the field, so state the conceptual payoff plainly. Inflating the novelty claim will not help.

Turning the excerpt into a revision plan

For each concern, write down the cause, the change that answers it, what the change costs in time, and where a reviewer will see it.

• Aim dependency and the null-result gap. Cause: Aim 2 assumes Aim 1 succeeds. Change: rewrite Aim 2 so it yields an interpretable result under either outcome, and add a short subsection on reading each aim if the effect is absent or small. Cost: a week and about a page. Location: Specific Aims and Approach.
• Unvalidated instrument. Cause: the measure was borrowed from a different population. Change: run a small validation study now and report it as preliminary data, or switch to an instrument with published validity in surgical patients. Cost: months, which is why it starts first. Location: one figure and one paragraph.
• Innovation ceiling. Cause: the application described what is new in the method and never said what it would teach about mechanism. Change: open the Significance and Innovation text with the mechanistic question. Cost: a careful rewrite of two pages. Location: the start of the Research Strategy.

Nothing on that plan touches the team or the environment, or the sections the reviewers praised. If every entry on your own plan is a wording fix, reread the bullets. Bullets that describe a design choice will not be answered by rewording the same design. The journal version of that mistake is covered in our guide to major revisions.

What the 2025 NIH review changes mean for an older A0

An A1 is reviewed against the criteria in the current funding opportunity, which may differ from those used for the A0 (NOT-OD-18-197). So an R01 submitted before January 25, 2025 and resubmitted now carries a critique organized under five scored criteria into a review scored on two factors. Map the old critique first.

• Significance and Innovation comments now feed one Factor 1 score. Innovation weaknesses can pull down the same number your Significance strengths hold up.
• Approach comments become Factor 2. The framework also moves inclusion considerations and the clinical trial study timeline into Factor 2 (NOT-OD-24-010), so answer those as Approach concerns.
• Investigator and Environment comments become a sufficiency question. If the old critique doubted the team's expertise in an area, answer with a named collaborator, a defined role and a letter.

The one-page Introduction

NIH changed the rules for this page in 2024 (NOT-OD-24-061). The Introduction must summarize substantial additions, deletions and changes, and respond to the weaknesses in the summary statement. It may not exceed one page unless the Table of Page Limits says otherwise. Changes may no longer be marked in the body of the application by any method, so the Introduction is the only place a reviewer learns what changed.

Here is a skeleton for the illustrative application.

Introduction

We thank the reviewers for their assessment. The committee agreed that the question is important and that the team is well placed to answer it. The score-driving concerns centered on rigor and feasibility, and we have revised the design in response.

Summary of substantial changes

- Aim 2 now produces an interpretable result whether or not Aim 1 finds an effect, and each aim includes a plan for null and small effects (Approach section [X]).

- New preliminary data validate the primary outcome measure in [N] surgical patients aged [AGE] and older (Figure [X]).

- The Significance and Innovation sections now open with the mechanistic question the adapted intervention answers.

Responses to weaknesses

Aim 2 depends on Aim 1 (summary bullets 2 and 4, Reviewer 2). [Two to four sentences. State the change, name the page, and stop.]

Outcome measure not validated in this population (summary bullet 3, Reviewer 2). [What you ran, what it showed, and where the figure is.]

Limited innovation (summary bullet 5, Reviewer 3). [Where you agree, say so. Then state the conceptual advance in one sentence and point to where the application develops it.]

Four habits keep it on one page and keep it persuasive.

• Paraphrase each concern in one line. Long quotations use up the space you need for answers.
• Cite the source. Naming the bullet or the reviewer shows you read the whole statement, and concerns from the bullets go first.
• Never argue with a score. Where a reviewer made a factual error, correct it once, with evidence, and then describe what you changed so the next reader cannot make the same mistake.
• End every response with a location. The reviewer should find each change without searching.

The same discipline applies to journal responses, which our response-to-reviewers template covers in more detail.

NIH rules that set your calendar

• One resubmission. NIH accepts a single A1 of an original application (NOT-OD-18-197).
• The 37-month window. After 37 months from the original submission, NIH treats any submission as a new application (NOT-OD-12-128).
• Life after a failed A1. You may submit the project again as a new A0, with no administrative requirement that the science differ. Renewals are the exception, since a renewal series allows only the A0 renewal and its A1 (NOT-OD-18-197).
• Calendar. For R01s, standard due dates are February 5, June 5 and October 5 for new applications and March 5, July 5 and November 5 for resubmissions. Other activity codes and many funding opportunities differ.

Because a failed A1 does not end the project, the A1 is less of a last chance than many PIs assume. The one-resubmission limit still argues against rushing it. If the longest item on your plan cannot finish before the next resubmission date, take the following cycle.

NSF: the revised proposal is the response

NSF treats a revised proposal as a new proposal under standard review (PAPPG 24-1, Chapter IV.E). Declined proposals may come back only after substantial revision, and one that has not clearly taken into account the major concerns from the prior review may be returned without review. Programs that accept proposals at any time may also set a moratorium on resubmitting a declined proposal.

Unless your solicitation asks for a separate response, the Project Description has to carry it. If you open with a short paragraph naming what changed, keep it brief, because it comes out of your page limit, and make every change it names visible in the sections that follow. Weight the two criteria equally, since a strong Intellectual Merit case does not rescue a Broader Impacts plan the panel called generic.

ERC: the interview is the rebuttal

ERC panels have a chair and 10 to 16 members who act as generalists, supported by remote referees who bring specialist expertise (ERC). The referee reports may raise technical doubts, and the people who ask about them in the interview may sit outside your subfield. Answer for a generalist first, then offer the detail.

• List the weaknesses you would raise yourself, and write a two-sentence answer with evidence for each.
• State the contribution in one sentence. If you cannot say plainly what the project will achieve that has not been achieved before, the panel will doubt the excellence claim.
• Answer the question you were asked. Panelists notice when a rehearsed block meets a different question.
• Spend any opening statement on the doubts. The panel has read the proposal.

If a proposal scored B, read the reapplication table in the current Work Programme before planning anything else. The ERC reserves the right to change those restrictions, and its Scientific Council readjusted the 2027 rules in April 2026.

Same program, a different program, or a different project

Resubmit to the same program when the concerns are specific and fixable within the design and the panel credited the question itself, as in the illustrative statement. Move programs when the doubts were about fit, and talk to the program officers in both places first. Rethink the project when two panels have returned similar criticisms, because that pattern points at substance. When reviewers on one panel disagree with each other, our guide to inconsistent reviews helps separate a real split from a difference in emphasis. For whether to keep going at all, see what to do after a rejection.

If you want a structured second reading of a critique before you start, ReviewPanel.ai can sort reviewer comments into causes and proposed changes alongside the proposal.

Frequently asked questions

Can you appeal an NIH or NSF grant review?

Both agencies have a process, and both are about procedure. At NIH, an appeal must describe a flaw in the review process, carry your institution's sign-off, and arrive within 30 days after the second level of review (NOT-OD-11-064). NIH will not review an A1 while an appeal on the A0 is pending (NOT-OD-11-101). At NSF, you can request reconsideration within 90 days of the declination, and it asks whether the review was fair and reasonable (PAPPG 24-1, Chapter IV.D). Disagreeing with the reviewers' scientific judgment is rarely a procedural flaw.

How many times can you resubmit an NIH grant?

Once. NIH accepts a single A1 per original application (NOT-OD-18-197), within 37 months of the original submission (NOT-OD-12-128). After an unsuccessful A1 you may submit the project as a new A0, and NIH does not require the science to differ. Renewal series allow only the A0 renewal and its A1.

Should I highlight changes in an NIH resubmission?

No. Since May 25, 2024, NIH guidance says not to mark changes in the body of a resubmission by any method. Describe them in the one-page Introduction, which must also respond to the weaknesses in the summary statement.

What does "competitive but not discussed" mean on an NIH summary statement?

It is a category NIH introduced for the 2026 councils and the January 2027 council (NOT-OD-26-114). Applications the committee votes into the middle third are not discussed and get no overall impact score, but they are still considered for funding. Treat the score-driving bullets on such a statement exactly as you would on a discussed application.

Do I have to resubmit to the same NSF program?

No. Fit varies between programs, so talk to both program officers first. Wherever it goes, a declined proposal must be substantially revised, and one that ignores the major concerns from its prior review can be returned without review.