Skip to content

NSF Grant Success Rate: FY2025 Funding Data

Updated 7 official sources

Quick Answer: If you are deciding whether to apply, NSF's overall grant funding rate was 19% in FY2025, down from 26% in FY2024.

Full answer and how we checked it

NSF decided 43,532 competitive proposal actions and made 8,377 new awards in FY2025. That equals about one new award for every 5.2 decided proposal actions, but it isn't your personal acceptance rate. Check the exact program, current solicitation, and recent awards before deciding whether to apply.

Last updated and source-checked: September 15, 2026.

Source review by Grantsights, September 15, 2026: We checked the FY2025 NSF Agency Financial Report, NSF's proposal and award process, NSF By the Numbers, current NSF Award Search, and the live BIO and MPS funding-rate pages. Several older directorate URLs now redirect to NSF By the Numbers, so this guide does not present a full current directorate table as though every row still had a live source.

Data note: FY2024 and FY2025 agency totals come from NSF's FY2025 Agency Financial Report. Directorate examples come from separate NSF pages and remain labeled FY2024. The proposal-actions-per-award figures are Grantsights calculations that divide competitive proposal actions by new awards and round to one decimal place. They are not NSF forecasts or applicant-specific probabilities.

In this guide

If you are a principal investigator, research administrator, or proposal lead deciding whether an NSF opportunity deserves a full application, treat the agency rate as the starting point and then narrow the comparison to your program. The phrase "NSF acceptance rate" usually points to NSF's official funding rate. The distinction matters because NSF counts proposal actions that reached a final award or decline decision in a fiscal year. It does not simply divide awards by every submission received that year.

This page answers three applicant questions:

  1. What is the latest agency-wide rate?
  2. What changed from FY2024 to FY2025?
  3. Which evidence should control an actual apply-or-skip decision?

From Manusights, our sister product

Writing an NSF proposal? Manusights offers an AI critique of your proposal against the review criteria in your solicitation: $149 for a research proposal, $49 for a GRFP or other fellowship application. Back within 3 business days of your upload.

Order a grant review

NSF funding rate at a glance

MeasureFY2024FY2025What changed
Overall funding rate26%19%Down 7 percentage points
Competitive proposal actions40,45443,532Up 8%
New awards10,5918,377Down about 21%
Average annual award size$256,099$260,362Up about 2%
Proposal actions per new award, derived3.85.2More decided actions per new award

Source: NSF FY2025 Agency Financial Report (opens in a new tab).

Based on our analysis of the FY2024 and FY2025 rows in the NSF FY2025 Agency Financial Report (opens in a new tab), the useful finding is the split between volume and awards. Competitive proposal actions increased by 3,078 while new awards decreased by 2,214. The average annual award size moved only about 2%. For an applicant, the useful conclusion is not that all NSF programs became equally harder. It is that the agency-wide rate became a weaker shortcut for program choice.

According to NSF's FY2025 Agency Financial Report, the funding rate fell 7 percentage points while the average annual award size rose about 2%. Those paired changes are why one agency percentage can't explain the outlook for a specific program.

The arithmetic also makes the change easier to see. In FY2024, 40,454 actions divided by 10,591 new awards equals about 3.8 actions per new award. In FY2025, 43,532 divided by 8,377 equals about 5.2. That 1.4-action difference describes the agency's decided portfolio. It does not mean every funded proposal beat exactly 4.2 competitors, because proposal volume and awards are not evenly distributed across programs.

Get research funding calls by email

About once a week, we email open research funding calls with a real deadline, from federal agencies such as NIH and NSF, picked for research institutions. One click to unsubscribe.

No spam. Unsubscribe anytime.

What NSF includes in the rate

NSF directorate pages define funding rate as:

new competitively reviewed awards divided by competitive awards and declines decided in the fiscal year

The decision year is important. A proposal is counted in the year NSF makes the final decision, not necessarily the year the proposal was submitted.

The public directorate definition excludes preliminary proposals, contracts, Intergovernmental Personnel Act agreements, supplements, continuing grant increments, graduate research fellowships, and similar actions. That makes the rate useful for broad competitive research context, but it does not describe every NSF funding path.

According to NSF, the agency strives to notify applicants within 6 months after submission whether a proposal has been declined or recommended for funding. That timing goal and the fiscal-year decision rule create another comparison problem: proposals submitted in one fiscal year can appear in the next year's funding-rate denominator. A one-year rate is therefore a record of decisions, not a same-year applicant cohort.

What NSF's agency report doesn't tell you

The FY2025 report gives 4 agency-level quantities that matter here: competitive proposal actions, new awards, funding rate, and average annual award size. It does not publish an FY2025 rate for every directorate, division, program, solicitation, institution type, career stage, or submission attempt in one comparable table.

That missing detail is the honest friction in success-rate research. A clean-looking chart can combine unlike units without making the mismatch obvious. Common mistakes include placing an FY2024 directorate rate beside an FY2025 agency rate, treating a division rate as a program rate, or using the number of awards returned by a keyword search as the denominator. Each produces a percentage, but none answers the same question.

The report also cannot tell you whether a proposed project is a strong fit. Scientific fit is evaluated against the program and solicitation, while the 19% rate pools decisions from many competitions. Nor can the report prove why the rate changed. It documents higher proposal actions and fewer new awards, but it does not assign one cause to every program.

Our September 2026 source review found another practical limit: some former directorate funding-rate URLs now redirect to NSF's broad By the Numbers page. We treat a redirect without the prior table as missing current evidence, not permission to repeat an old percentage. That is why this page shows only the BIO and MPS examples that still have direct source pages.

Which rate should you use?

Use this order instead of applying the 19% agency number to every proposal.

DecisionEvidence to checkWhy it changes the answer
Is NSF the right funder?NSF funding search (opens in a new tab)Confirms that the subject and applicant type belong at NSF.
Is this the right program?Active solicitation and program pageControls scope, eligibility, proposal limits, dates, and review details.
Is this the right division?Directorate or division rate when NSF publishes oneGives a closer historical denominator than the agency average.
Is the project scale plausible?NSF Award Search (opens in a new tab)Shows recent titles, institutions, amounts, and program ownership.
Is the proposal ready?NSF proposal and award process (opens in a new tab)Confirms registration, submission, review, and decision steps.

If scientific fit is weak, a higher historical directorate rate does not repair it. If the program fit is strong, a lower rate does not automatically mean the proposal should be abandoned. The rate is a capacity signal, not a routing rule.

Use a 3-level comparison whenever the sources allow it:

  1. Agency context: the FY2025 19% rate answers what happened across NSF.
  2. Directorate or division context: a published rate can show whether a closer scientific portfolio differed in a named fiscal year.
  3. Program evidence: the active solicitation and recent awards show what applicants, topics, budgets, and program elements belong in the actual competition.

Stop at the level the evidence supports. If NSF publishes no current program denominator, label the gap. Do not manufacture a rate from search results because Award Search may include continuing actions, supplements, or records that do not match the official funding-rate denominator.

Directorate rates need source labels

NSF still publishes some directorate and division tables. The live BIO funding-rate page (opens in a new tab) reports a 19% overall BIO rate for FY2024, with division rates of 13% for DBI, 24% for DEB, 19% for IOS, and 23% for MCB. The live MPS funding-rate page (opens in a new tab) reports a 27% MPS rate for FY2024.

These rows are historical FY2024 measures. Do not mix them with the FY2025 agency rate and call the result a current directorate ranking. Also do not move a project from BIO to MPS because 27% is higher than 19%. The scientific question, method, program scope, and review community must support the route.

The BIO table shows why a closer comparison can change interpretation. BIO's FY2024 directorate rate was 19%, but its listed divisions ranged from 13% to 24%, an 11-point span. Even that division view is still broader than one solicitation. According to NSF's BIO data, the division rows use the same final-decision-year concept, which makes them internally comparable for that year but not predictive of the next competition.

A worked decision example

Consider one investigator at a university choosing between a BIO core program and an interdisciplinary opportunity. This is an illustrative workflow, not a claim about a named applicant. The project clearly fits biological infrastructure, and the investigator sees BIO's 19% FY2024 rate. The agency's FY2025 rate is also 19%, so the two percentages appear to confirm each other.

The percentages alone still do not decide the route. The investigator should take 4 additional steps:

  1. Read both active solicitations and mark the required program elements.
  2. Search each program element in NSF Award Search, rather than searching only a broad scientific keyword.
  3. Sample at least 10 recent awards when the program has enough records, recording title, abstract, institution, amount, start date, and program element.
  4. Ask the program contact a concise scope question if the scientific ownership remains unclear.

Suppose the core program's recent awards repeatedly match the project's method and research question, while the interdisciplinary route's awards require a partnership or application setting the team does not have. The core program is the better route even if the other opportunity seems to have a higher historical percentage. The decision changed because of fit evidence, not because the team converted 19% into personal odds.

This workflow also exposes a bad-fit proposal early. A team may discover that its project is mainly product development, education delivery, or deployment rather than research within the named program. The next action is then to find a better route, not to polish the same abstract around an attractive acceptance-rate number.

A five-step apply-or-skip screen

1. Confirm the current solicitation

Use the solicitation named on the program page. Check applicant type, proposal limits, deadline or window, required documents, and any program-specific review criteria.

2. Read recent award abstracts

Search the exact program in NSF Award Search. Review recent project titles, abstracts, institutions, award amounts, start dates, and program elements. This tests whether the proposed work resembles what the program actually funds.

Use a small evidence sheet with 6 columns: award number, title, institution, amount, start date, and fit note. Record why each award is relevant instead of copying keywords. Ten well-matched records can reveal program ownership more clearly than 100 broad keyword results, although neither set creates an official success rate.

3. Compare the budget with real awards

An agency average does not set a proposal budget. Use the solicitation ceiling and a set of recent awards in the same program. Explain any material difference between your request and the observed range.

For a multi-year award, compare the total intended amount and annualized amount separately when the record supports both. A $1 million, 5-year award and a $1 million, 2-year award create different annual spending patterns. Do not infer an award ceiling from the largest search result; the solicitation controls.

4. Test the program fit before a full draft

NSF's process page advises applicants to use program contacts and the active funding opportunity. A short fit discussion can identify the right program or expose a scope mismatch before the team spends weeks drafting.

Send a short, answerable question: name the research objective, method, applicant institution, and the program element you believe owns the work. Do not ask a program officer to pre-review a full proposal or promise competitiveness. The aim is route clarity.

5. Treat rate data as a portfolio input

Use the rate to plan workload and risk across a research portfolio. Do not turn it into a promise for one proposal. A team can pair one high-effort core proposal with a different eligible route, but each submission still needs its own source-backed fit case.

For portfolio planning, separate 3 judgments: fit, readiness, and capacity. A strong-fit project can still be unready because pilot data, partnerships, registrations, or required documents are missing. A ready project can still be a poor use of capacity if the team cannot finish a compliant application by the deadline. The rate informs the risk discussion only after those checks.

What the first-attempt data does and does not prove

NSF's proposal-process page reports two historical facts: PIs submitted an average of 2.3 proposals for every award, and 36% of new PIs who received a first NSF award in 2020 received it on their first attempt.

Those facts do not prove that resubmission itself causes success. Later proposals may have a different program, scope, team, budget, solicitation, or review context. Use feedback to improve the next submission, but rebuild the fit check against the current solicitation.

According to NSF's process overview, 36% of new PIs who received their first NSF award in 2020 did so on the first attempt. That sentence has two conditions that are often dropped: it describes new PIs who ultimately received a first award, and it describes 2020. It is not the percentage of all first-time applicants funded on their first attempt.

The 2.3-proposals-per-award figure has a similar limit. NSF presents it as an average historical relationship. It does not distinguish a revised version of the same proposal from a different proposal to another program, so it cannot support a causal claim that resubmission raises an individual proposal's odds.

Build your own program evidence sheet

A 10-award sample will not produce an official acceptance rate, but it can make program fit testable. Create one row per recent award and capture these fields:

FieldWhat to recordDecision use
Program elementExact NSF labelConfirms program ownership
Award start dateMonth and yearKeeps the sample current
Institution and PIRecipient shown by NSFReveals the funded applicant set without inventing eligibility
Title and abstractOne-sentence research focusTests topic and method fit
Total intended amountValue shown in the recordCreates a sourced scale comparison
DurationStart and estimated endSeparates total amount from annual pace
Fit noteStrong, partial, or weak, with one reasonForces an explicit route judgment

Then summarize the sample without converting it into a success rate. For example: "7 of 10 reviewed awards used the program element named in our draft, 6 involved the same core method, and 2 had a similar institutional role." That is a description of the selected sample, not the odds of funding.

Check the sample against the solicitation after collecting it. An award made under an older solicitation may show durable scientific ownership, but current eligibility, budget, documents, and review rules still come from the active notice. If the source record lacks a field, leave it blank rather than filling the gap from an unrelated page.

For a specialized route, use a specialized guide after the agency-level screen. The NSF EPSCoR guide explains the jurisdiction and capacity-building route, while the NSF REU guide covers undergraduate research-site and supplement choices. Neither page should inherit the 19% agency rate as its program odds.

If the project needs research equipment, the NSF MRI equipment guide adds the route-specific institution, instrument, and proposal-limit checks that an agency success rate doesn't show.

Common errors when reading NSF success rates

  • Treating 19% as personal odds. The agency denominator contains many programs and proposal types.
  • Calling every NSF action an application. The published rate excludes several action types.
  • Using the decision year as the submission year. NSF assigns the action to the fiscal year of final decision.
  • Choosing a directorate by percentage. Scientific and program fit come first.
  • Budgeting from an agency average. The solicitation and recent program awards are better evidence.
  • Claiming that a later proposal has guaranteed better odds. NSF does not publish a universal causal resubmission rate.
  • Building a denominator from Award Search. Search results are useful for fit and scale, but they are not automatically the official set of competitive decisions.
  • Dropping the fiscal year. A 19% figure without FY2025 can be mistaken for a standing rate.
  • Comparing unlike award amounts. Total intended amount and average annual award size answer different questions.

The biggest error is false precision. Writing that a specific applicant has a 19% chance may sound analytical, but the public data does not support it. A defensible conclusion names the source, fiscal year, unit, and limit, then explains what evidence should decide the route.

Where to go next

Sources

Download this page's tables (CSV). When you use the numbers, credit Grantsights and the source named on the page; see how to cite.

More award history and success-rate data

Who won, how many awards were made, and success rates, from public award records and agency tables.

NIH

USDA NIFA

SBIR

Frequently Asked Questions

What is the current NSF grant success rate?

NSF reported an overall funding rate of 19% for FY2025, down from 26% in FY2024. NSF calculates this as new competitive awards divided by competitive proposal actions decided during the fiscal year. It is not an applicant-specific probability.

Why did the NSF funding rate fall in FY2025?

NSF's FY2025 Agency Financial Report shows competitive proposal actions rose 8%, from 40,454 to 43,532, while new awards fell from 10,591 to 8,377. The public totals show both more decided proposals and fewer new awards, but they do not establish a program-level cause.

Does NSF acceptance rate mean the same thing as funding rate?

People often say acceptance rate, but NSF uses funding rate. Its denominator is competitive proposal actions that reached an award or decline decision. It excludes preliminary proposals, supplements, continuing grant increments, graduate research fellowships, contracts, and several other actions.

Can I use the 19% rate to estimate my proposal's odds?

Not reliably. Directorate, division, program, solicitation, proposal type, available funding, and scientific fit can all change the comparison set. Use 19% as agency context, then inspect the active solicitation and recent awards in the exact program.

Does resubmitting an NSF proposal improve the success rate?

NSF does not publish a universal causal success-rate advantage for later submissions. NSF reports that PIs averaged 2.3 proposals per award and that 36% of new PIs receiving a first award in 2020 did so on their first attempt. Those facts describe history, not a guarantee for a revised proposal.

How long does NSF take to decide?

NSF says it strives to tell applicants within six months whether a proposal has been declined or recommended for funding. Program timing can vary.

Last updated September 15, 2026. Published by Grantsights from public sources. See our editorial policy, how we build and check guides, corrections and how we make money.