National Science Foundation Rejections: Why 78% of NSF Proposals Fail (And What Funded PIs Do Differently)

The National Science Foundation Rejects Most Proposals. Here Are the Numbers.

If your National Science Foundation proposal was rejected, the first thing to understand is that rejection is the default outcome. The NSF funds between 22% and 28% of proposals in a typical year. In FY 2025, the agency received over 42,000 proposals and declined roughly 32,000 of them.

That is not a reflection of proposal quality alone. The NSF's budget has remained relatively flat in real dollars while submission volume has increased steadily. The result is a funding environment where technically sound, well-written proposals routinely get rejected because there simply is not enough money to fund them all.

But that does not mean rejection is random. Certain patterns emerge consistently in reviewer feedback, and PIs who get funded tend to do specific things differently from those who do not.

NSF Rejection Rates by Directorate

Not all parts of the NSF are equally competitive. Funding rates vary significantly by directorate:

SBE and BIO are consistently the most competitive directorates. If you are submitting to one of these, a rejection does not necessarily mean your proposal was weak — it may mean you were competing against an unusually strong pool.

The 6 Reasons NSF Proposals Actually Get Rejected

After reviewing hundreds of NSF panel summaries and working with PIs across every directorate, these are the rejection reasons that come up repeatedly. They are listed in order of how frequently they appear in reviewer feedback.

1. Weak or Missing Broader Impacts

This is the single most common critique in declined NSF proposals. Many PIs treat the broader impacts section as an afterthought — a paragraph about mentoring undergrads tacked onto the end of an otherwise strong research plan.

NSF reviewers score broader impacts as a co-equal criterion with intellectual merit. A proposal with outstanding science and generic broader impacts will lose to a proposal with strong science and genuinely thoughtful broader impacts.

What funded PIs do differently: They integrate broader impacts into the research design itself rather than bolting it on as a separate section. A computational biology PI who builds her analysis tools as open-source packages with documentation and tutorials for community college instructors is demonstrating broader impacts through the work itself — not in addition to it.

2. Unclear or Overly Ambitious Research Plan

Reviewers need to believe your project is both important and feasible within the proposed timeline and budget. The most common failure mode is a proposal that tries to do too much.

Three well-defined aims with clear milestones will always score better than five ambitious aims where any one of them could fill an entire grant period.

What funded PIs do differently: They scope ruthlessly. Each aim has a concrete deliverable, a clear methodology, and an explanation of what happens if a particular approach does not work. Reviewers want to see that you have thought about contingencies, not just the optimistic path.

3. Insufficient Preliminary Data

The NSF does not expect completed research in a proposal. But reviewers need evidence that your approach is feasible. This is especially true for CAREER awards and proposals involving novel methods.

The gap is usually not about the amount of preliminary data but about how it is presented. A single well-designed pilot experiment with clear results and an honest discussion of limitations is more persuasive than a wall of figures without interpretation.

What funded PIs do differently: They connect every piece of preliminary data to a specific aim. If your pilot work tested approach X and you are proposing to use approach Y, you need to explain why — otherwise reviewers will question whether the preliminary data actually supports the proposed work.

4. Poor Alignment with the Program Solicitation

Submitting a strong proposal to the wrong program is one of the most avoidable mistakes. Each NSF program has specific priorities, and proposals that do not address those priorities will be declined regardless of scientific quality.

What funded PIs do differently: They contact the program officer before submitting. A 15-minute conversation with a PO can tell you whether your project fits, which program is the best match, and whether there are specific priorities for the current cycle that your proposal should address. This is free, it is expected, and most PIs who get funded do it.

5. Writing That Does Not Communicate to a Broad Panel

NSF review panels are interdisciplinary. Your proposal will be read by experts in your subfield and by scientists in adjacent areas who may not share your technical vocabulary.

Proposals that assume too much domain knowledge from the reader will score poorly on clarity even if the science is excellent. This is different from dumbing down your work — it is about making your argument accessible without sacrificing precision.

What funded PIs do differently: They have someone outside their immediate subfield read the proposal before submission. If that person cannot explain the main idea back to you in two sentences, the proposal is not clear enough.

6. Budget or Administrative Issues

This is the least interesting rejection reason but it still accounts for a meaningful share of declined proposals. Common issues include budgets that do not match the scope of work, missing required sections (data management plan, mentoring plan for postdocs), and page limit violations.

What funded PIs do differently: They use a checklist. The NSF Proposal & Award Policies & Procedures Guide (PAPPG) changes regularly, and requirements that applied to your last submission may have changed. A systematic check against the current PAPPG catches problems that are embarrassing to miss.

What to Do After an NSF Rejection

Read the Panel Summary Like a Reviewer

The panel summary is the most valuable document you will receive from the NSF. It synthesizes the individual reviews into a consensus view of your proposal's strengths and weaknesses.

Do not read it the day you receive it. Give yourself a few days, then read it with a pen and mark every specific critique. Categorize each one:

• Fixable in revision — unclear writing, missing preliminary data, budget justification issues
• Requires rethinking — scope too broad, wrong program fit, fundamental methodology concerns
• Disagreements — places where you believe the reviewer misunderstood your proposal (these matter too, because if the reviewer misunderstood, the writing was not clear enough)

Talk to the Program Officer

After every declined proposal, you can (and should) request a debrief with the program officer. They can provide context that the written reviews do not capture — how the panel ranked your proposal relative to others, whether the concerns were unanimous or split, and whether a resubmission to the same program makes sense.

Some program officers will tell you directly whether a resubmission is worth your time. That information alone makes the call worthwhile.

Decide Whether to Resubmit or Pivot

Not every rejected proposal should be resubmitted to the same program. If the reviews suggest fundamental concerns about the approach or the fit, consider:

• Submitting to a different NSF program that better matches your work
• Reframing the project for a different funding agency (NIH, DOE, DOD all fund research that overlaps with NSF's scope)
• Splitting an overly ambitious proposal into two focused proposals

If the reviews are generally positive with specific, addressable concerns, a resubmission is usually the right move. NSF does not penalize resubmissions, and reviewers tend to look favorably on proposals that demonstrate responsiveness to feedback.

Strengthen the Proposal Systematically

When resubmitting, do not just patch the specific issues reviewers raised. Use the feedback as a starting point to reassess the entire proposal:

• Rewrite the project summary and broader impacts section from scratch
• Ask two colleagues outside your subfield to review the revised draft
• Verify alignment with any new program priorities or solicitations
• Update your preliminary data with any results generated since the original submission

NSF Rejection Is Normal. Giving Up Is Not.

The median funded PI at the NSF has been rejected at least once before getting funded. Many have been rejected multiple times. The difference between PIs who eventually get funded and those who do not is rarely about talent or the quality of their ideas. It is about persistence, responsiveness to feedback, and the willingness to revise seriously rather than superficially.

If you are working on your next NSF proposal — whether it is a first submission or a resubmission after a rejection — tools that help you systematically address reviewer criteria can make the process more efficient. Granted AI analyzes your solicitation document and tracks whether your draft addresses each requirement, so you can identify gaps before reviewers do.

Related reading:

• What NSF Reviewers Wish You Knew
• NSF Broader Impacts: Examples That Actually Work
• NSF Grant Writing Guide: From Concept to Submission