NSF’s agency-wide funding rate for competitively reviewed proposals has held near 27% in NSF’s own recent Funding Profile estimates — but that single number hides enormous variation. Within a single fiscal year, division-level funding rates at NSF have ranged from single digits to over 40%, depending on directorate, division, and how many proposals a given panel received relative to its budget. Anyone advising a principal investigator on “how hard is it to get NSF funding” needs both numbers: the agency-wide baseline, and the specific directorate/division rate for the program actually being targeted.
This guide walks through NSF’s own published funding-rate data, explains exactly what “funding rate” means as NSF defines and calculates it, shows why the number varies so much by field, and points to where research administrators can find the current figure for a specific program rather than relying on a stale or agency-wide average.
What NSF’s “funding rate” actually measures
NSF defines the funding rate (sometimes called the “success rate”) as the number of awards made during a fiscal year as a percentage of total proposals competitively reviewed in that year. NSF itself is explicit that this is a probability estimate, not a guarantee: it “indicates the probability of receiving an award when submitting proposals to NSF,” calculated after the fact from actual award and decline counts, not a target or quota set in advance.
Two distinctions matter when reading NSF’s figures correctly:
- Competitive awards vs. research grant awards. NSF publishes two related but different figures. “Competitive awards” is the broader category — it includes research grants, cooperative agreements, equipment grants, fellowships, and conferences. “Research grant awards” is the narrower subset limited to research projects, excluding those other award types. The research-grant-only rate is consistently a few points lower than the all-competitive-awards rate, because fellowship and conference actions tend to have different (often higher) funding rates that pull the blended number up.
- Funding rate is not the same as an individual PI’s odds. The published rate is a proposal-level statistic across an entire program or directorate in a given year, not a per-applicant probability. It doesn’t account for resubmissions (a declined proposal that comes back revised counts as a new proposal action), doesn’t distinguish new PIs from established ones, and says nothing about a specific panel’s competitiveness in a specific cycle.
NSF’s overall funding rate: the numbers
The most authoritative source for the agency-wide figure is NSF’s own Funding Profile, published as part of its annual Budget Request to Congress. The FY 2025 Budget Request’s Funding Profile table reports:
| Statistic | FY 2023 (base plan estimate) | FY 2025 (request estimate) |
|---|---|---|
| Competitive proposals | 42,000 | 40,800 |
| Competitive awards | 11,200 | 10,900 |
| Funding rate, all competitive awards | 27% | 27% |
| Research grant proposals | 37,000 | 36,200 |
| Research grant awards | 9,400 | 8,900 |
| Funding rate, research grants only | 25% | 24% |
| Median annualized award size | $181,900 | $193,700 |
| Average award duration | 3.1 years | 2.5 years |
Read together, NSF’s own baseline has sat consistently in the mid-to-high 20s for the all-competitive-awards rate and the mid-20s for research grants specifically across these estimates. That is the honest answer to “how hard is it to get NSF funding” at the whole-agency level: roughly three out of every four competitively reviewed proposals do not get funded in a given cycle. Budget Request figures are planning estimates prepared before a fiscal year closes, so treat them as the agency’s own best estimate rather than a final audited actual, and expect the number to move a point or two from what’s ultimately realized once a year’s competitions are complete.
Why the rate varies so much by directorate and division
The agency-wide number is an average across seven research directorates, and the underlying rates diverge sharply. NSF publishes funding rates for each directorate and its constituent divisions annually. For fiscal year 2024, directorate-level funding rates included:
| Directorate | FY 2024 funding rate |
|---|---|
| Geosciences (GEO) | ~36% |
| Mathematical and Physical Sciences (MPS) | 27% |
| Engineering (ENG) | 23% |
| Computer and Information Science and Engineering (CISE) | ~22% |
The spread gets wider still one level down, at the division within each directorate. Within MPS’s FY 2024 data, for example, Physics awarded 45% of reviewed proposals (285 of 627 actions) while Astronomical Sciences awarded 18% (184 of 1,027 actions) — both inside the same directorate, in the same fiscal year. Within ENG, the Emerging Frontiers and Multidisciplinary Activities (EFMA) division funded just 8% of actions (24 of 287) while Environmental Engineering and Education (EEC) funded 33% (111 of 333). A PI in a low-funding-rate division is competing in a genuinely different environment than a PI in a high-funding-rate division of the same directorate, even though both would describe themselves simply as “NSF-funded.”
A few structural reasons drive this spread, beyond simple randomness:
- Proposal pressure differs by field. Some divisions receive far more proposals per dollar of available budget than others, independent of scientific merit — a crowded subfield with a fixed appropriation produces a lower rate almost mechanically.
- Congressional appropriations set directorate-level budgets, and those allocations don’t move in lockstep with proposal volume from year to year. A directorate whose appropriation grows more slowly than its proposal pool will see its funding rate fall even if proposal quality is unchanged. See how NSF’s budget and appropriations process affects grant award timing for how that mechanism works.
- Program-specific solicitations and Dear Colleague Letters can spike proposal volume into a narrow competition for a fixed pot of funds, temporarily depressing that specific competition’s rate without reflecting the directorate’s typical rate.
- Award size and duration trade off against count. A division that funds fewer, larger, longer awards will show a lower funding rate for the same total budget than one funding many smaller, shorter awards.
Because of this spread, an agency-wide “27%” is a poor proxy for any individual PI’s actual odds. The number that matters for planning a submission is the rate for the specific division or program the proposal will be reviewed in, not the NSF-wide average.
Where to find the current rate for a specific program
NSF publishes directorate- and division-level funding rate tables on its own site, updated annually, at pages such as nsf.gov/<directorate>/funding-rates (for example, the Mathematical and Physical Sciences and Engineering directorate pages). These are the authoritative source and should always be checked directly rather than relying on a figure quoted secondhand, since rates are recalculated every fiscal year as final award and decline counts are confirmed. For a broader orientation to where funding statistics live across agencies — not just NSF — see Research Grant Funding Statistics: Where to Look, which covers NSF’s National Center for Science and Engineering Statistics (NCSES) resources alongside NIH RePORT and UKRI data.
For general NSF proposal mechanics that affect what actually gets counted in these statistics — solicitation types, review criteria, submission requirements — see NSF Grants: An Overview, and for how proposals are actually scored before the funding-rate math is applied, see the NSF Merit Review Panel dictionary entry.
NSF vs. NIH: why the numbers aren’t directly comparable
Research administrators who work across both agencies often want to compare NSF’s funding rate to NIH’s, but the two agencies report competitiveness in structurally different ways. NSF reports a funding rate: awards divided by competitively reviewed proposals, published by directorate and division. NIH instead centers its competitiveness reporting around the payline — a percentile cutoff on peer-review scores that a given institute sets for a funding cycle, which functions differently from a simple award/proposal ratio and varies by institute and mechanism rather than by broad directorate. See NIH Payline for how that concept works, and NIH vs. NSF Resubmission Policy for a related structural difference between the two agencies that also affects how “success” gets counted across resubmissions. Because the underlying metrics aren’t calculated the same way, a bare comparison of “NSF is X%, NIH is Y%” is misleading without accounting for that difference in methodology — and for agencies where research administrators split effort across both, NIH vs. NSF Data Management Plans covers a related area where the two agencies’ requirements diverge in ways that matter for proposal preparation, not just funding odds.
What this means in practice for research administrators
- Don’t quote the agency-wide rate as if it applies to a specific proposal. Pull the directorate and, where available, division-level rate for the program the PI is actually targeting before setting expectations.
- Treat the rate as a planning input, not a forecast. A 20-something percent funding rate means most well-reviewed proposals are still declined in any given cycle; portfolio strategy (multiple funding sources, staged submissions, resubmission planning) matters more than optimizing around a single agency’s published rate.
- Watch appropriations timing. Because directorate budgets are set annually and proposal pools shift year to year, a rate from two or three years ago can be a meaningfully worse predictor than the most recently published figure — check the current-year table directly rather than reusing a number from an older briefing deck.
- Distinguish “funding rate” from “success rate” in casual conversation carefully. NSF uses the two terms interchangeably for the same competitive-awards calculation, but faculty and administrators sometimes use “success rate” loosely to mean an individual PI’s personal track record, which is a different (and non-public) statistic.
Frequently asked questions
What is NSF’s overall funding rate?
Per NSF’s own Funding Profile in its FY 2025 Budget Request to Congress, the agency-wide funding rate for all competitively reviewed proposals was estimated at 27% for both FY 2023 and the FY 2025 request estimate. The narrower research-grant-only rate was 25% (FY 2023 estimate) and 24% (FY 2025 estimate). These are agency-wide averages; directorate and division rates vary well outside this range in both directions.
Is a 27% funding rate considered low?
It means roughly three in four competitively reviewed proposals are not funded in a given year, which is broadly comparable to the general order of magnitude reported for other major federal research funders, though the exact figures and how they’re calculated differ by agency. Whether 27% counts as “low” depends heavily on the comparison point and, more usefully for planning purposes, on the specific division’s rate rather than the agency-wide figure.
Does the funding rate differ by NSF directorate?
Yes, substantially. NSF’s own FY 2024 directorate-level data shows rates ranging from roughly 22% (Computer and Information Science and Engineering) to roughly 36% (Geosciences), and division-level rates within a single directorate can range even more widely — for example, from 8% to 33% among Engineering’s divisions in the same fiscal year.
Does resubmitting a declined NSF proposal improve the odds?
NSF’s funding rate statistics count each competitively reviewed proposal action separately, including resubmissions, rather than tracking an individual proposal’s history through multiple review cycles. NSF does not publish a separate “resubmission success rate” in its Funding Profile data. Program officer feedback on the original decline, and whether the reviewers’ specific concerns were substantively addressed, are the practical factors that matter for a resubmission’s chances, not a generic resubmission bonus baked into the published statistics.
Where can I find the current funding rate for a specific NSF program?
Check the relevant directorate’s own funding-rates page on nsf.gov directly (each of NSF’s directorates publishes and updates its own table annually), rather than relying on an agency-wide figure or a number from a prior fiscal year. See the “Where to find the current rate” section above for how these are organized.
Sources
- National Science Foundation, FY 2025 Budget Request to Congress, “NSF Funding Profile” summary table (nsf-gov-resources.nsf.gov).
- National Science Foundation, Directorate for Mathematical and Physical Sciences, “MPS Funding Rates” (nsf.gov/mps/funding-rates).
- National Science Foundation, Directorate for Engineering, “ENG Funding Rates” (nsf.gov/eng/funding-rates).
- National Science Foundation, directorate-level funding rate pages for Geosciences (GEO) and Computer and Information Science and Engineering (CISE) (nsf.gov/funding/funding-rates.jsp).
Figures in this guide reflect the most recently published NSF data available at the time of writing (FY 2023 estimates, FY 2024 directorate actuals, and FY 2025 request estimates). NSF recalculates and republishes these figures annually; always confirm against NSF’s current directorate pages before using a figure in a proposal-planning conversation.
