The NSF SBIR Phase I success rate is roughly 14-20%, making NSF the highest-funded agency among all 11 SBIR participants. But that headline number hides a funnel. Roughly 50-65% of Project Pitches get invited (NSF doesn't publish official rates), 14-20% of full proposals get funded, and about 34% of Phase I awardees go on to win Phase II. End-to-end from first pitch to Phase II money, the real rate is 3-5%.
Most content on this topic quotes one number and stops. That's not useful for making a decision. You need the full funnel, the cross-agency comparison, and the levers that actually move your odds.
The full NSF SBIR funnel
Here's what happens to 1,000 companies that submit a Project Pitch:
| Stage | What Happens | Approximate Rate | Companies Remaining |
|---|---|---|---|
| Project Pitch | Submit ~1,500-word screening form | ~50-65% invited | ~500-650 invited |
| Full Proposal | Submit 15-page technical proposal | ~14-20% funded | ~90-120 win Phase I |
| Phase I Execution | Complete 6-12 month R&D project | ~80% apply for Phase II | ~72-96 apply |
| Phase II Proposal | Submit Phase II proposal | ~43% funded | ~31-41 win Phase II |
End-to-end from pitch submission to Phase I award: ~9-12%. From pitch to Phase II award: ~3-5%.
Those numbers might seem discouraging. But compare them to what you're actually risking at each stage:
- Project Pitch: 3-4 hours of work. If declined, you've lost one afternoon and gained specific feedback.
- Full Proposal: 80-120 hours. This is where the real investment starts. But you only get here after a positive pitch signal.
- Phase II: You're building on Phase I data. The 43% rate is strong for a grant this size ($1.25M).
The funnel is designed to save you time. The pitch stage means you don't invest 100+ hours without a fit signal first.
How NSF compares to other agencies
NSF has the highest Phase I success rate of any SBIR agency. That matters for your decision about where to apply.
| Agency | Phase I Success Rate | Awards/Year | Phase I Amount | Key Difference |
|---|---|---|---|---|
| NSF | 14-20% | ~280 | $305,000 | Highest rate, most first-time-friendly |
| NIH | ~18% average (13-24% by Institute) | Varies by IC | Up to $314,363 | Preliminary data critical; institute-specific |
| DoD (all branches) | ~15% | ~1,440 | $50K-$250K | Largest volume; AFWERX is different model |
| DOE | ~13% | Smaller | $200K-$250K | Topic-specific; lower rate |
| Overall (all agencies) | ~17% | ~4,000+ | Varies | $4B+ annually |
NSF also has the least concentrated awards of any SBIR agency -- the fewest going to repeat SBIR companies with 10+ awards. That's intentional. NSF actively distributes funding broadly, which makes it more accessible for first-time applicants who don't have a track record of prior awards.
For the full agency-by-agency breakdown, see our SBIR funding amounts reference.
What actually moves the odds
The 14-20% headline rate is an average across all applicants. Your individual odds depend on several specific factors.
I-Corps alumni: 3x the baseline
This is the single biggest lever. Per NSF program data, companies that complete NSF I-Corps -- a free, 7-week customer discovery cohort -- have roughly a 50% SBIR success rate, about 3x the baseline. Since 2011, I-Corps alumni have launched 1,000+ startups and raised $760M+ in follow-on funding. Some of that gap is selection effects (teams motivated enough to do I-Corps were probably going to apply seriously anyway), but the correlation is strong enough that NSF's own Program Directors actively encourage it.
The mechanism is straightforward: I-Corps trains you in customer discovery and commercial validation, which directly strengthens the commercialization criterion NSF reviewers score. It also signals to the Program Director that you understand the market side, not just the science.
If you're planning to apply to NSF SBIR and have 7 weeks before your proposal deadline, completing I-Corps (or I-Corps Hub equivalent) may be the highest-ROI investment you can make.
The 20% compliance filter
An estimated 20%+ of SBIR applications across all agencies are rejected before review for administrative non-compliance -- wrong format, missing documents, exceeded page limits, eligibility issues. These proposals never reach a reviewer.
If you account for this, the "effective" success rate for properly submitted proposals is higher than the headline -- closer to 18-25% at NSF. The fix is straightforward: before submitting, walk through the solicitation's "Proposal Preparation Instructions" section line by line and verify every format requirement, page limit, and required document. The NSF Phase I solicitation spells out every requirement explicitly.
First-time vs. repeat applicants
First-time applicants at NSF are estimated to succeed at 13-15% -- only slightly below the overall average. That gap is smaller at NSF than at other agencies, specifically because NSF discourages repeat SBIR mills.
Contrast this with NIH, where preliminary data is close to a requirement, or DoD, where prior relationships with program managers can give repeat applicants a significant advantage. If you've never applied for a grant before, NSF is the most level playing field.
Technology area
NSF is technology-agnostic. Unlike DoD (which publishes specific topics) or NIH (which scores within Institute-specific study sections), NSF funds across nearly all science and engineering domains. Your technology area doesn't determine your odds at NSF -- the quality of your research question and commercial framing does.
NSF funds roughly 400 companies per year across AI, biotech, clean energy, advanced materials, cybersecurity, robotics, ag-tech, ed-tech, and more.
Resubmission
There's no published data on NSF SBIR resubmission success rates specifically. But the process allows it: if your pitch is declined, you can revise and resubmit at any time (rolling). If your full proposal is declined, you can submit a new proposal to the next cycle.
The key: resubmission requires real changes, not surface edits. You'll submit a "Resubmission Change Description" explaining how you addressed reviewer feedback. If the feedback was about framing (too product-focused, scope too broad), that's fixable. If it was about fundamental fit, consider a different agency.
For tactical advice on resubmission, see our post on interpreting SBIR feedback.
Why NSF SBIR success rates have declined
NSF SBIR success rates have trended slightly downward over the past decade. The cause isn't declining quality -- it's increasing award size.
When Phase I awards were $150K, NSF could fund more of them. As awards increased to $225K and then to $305K (current), the number of total awards decreased while applications grew. The result: the same budget funds fewer companies at higher amounts.
This is a trade-off, not a problem. Each Phase I winner gets substantially more money to prove feasibility. But it means the competition for each slot is tighter than it was five years ago.
The Phase I to Phase II gap
One number that doesn't get enough attention: only about 34% of Phase I awardees ultimately win Phase II.
Here's why:
- ~80% of Phase I awardees apply for Phase II (some don't finish Phase I, pivot, or raise enough private capital to not need it)
- ~43% of Phase II proposals are funded
Phase II proposals are evaluated more rigorously than Phase I. You need actual results from your Phase I work -- data showing feasibility, not just a plan. The reviewers are looking for evidence that Phase I de-risked the core technical question and that the Phase II plan builds credibly on those results.
If you're in Phase I now, start thinking about Phase II from day one. The strongest Phase II proposals are designed backwards from Phase I.
What this means for your decision
If you're weighing whether NSF SBIR is worth your time, here's how to think about the numbers:
The pitch is almost always worth it. 3-4 hours for a 60% chance of invitation and specific feedback either way. The expected value is high even if you think your odds are below average.
The full proposal is a bigger bet. 80-120 hours for a 14-20% chance of $305K. That's a good expected value if $305K in non-dilutive funding would meaningfully change your trajectory. It's a harder call if you're a 2-person team with 3 months of runway.
I-Corps changes the math. If you can complete I-Corps before applying, your success rate roughly triples. Seven weeks and a few thousand dollars of program cost for 3x odds on a $305K grant is an exceptional deal.
NSF is the right starting point for most first-time applicants. Highest success rate, broadest technology scope, most accessible two-step process, least biased toward repeat applicants. If you've never applied for SBIR and you're not sure where to start, start here.
For a step-by-step guide on writing your pitch, see our NSF SBIR Project Pitch guide. For the broader question of whether grants are right for your company, see should your startup apply for government grants.
Want to know your realistic odds?
Submit a Project Pitch draft through our Pitch First service. We'll tell you in 48 hours if the framing will get you invited -- before you invest 100+ hours on a full proposal. Or if you're still deciding whether NSF is the right agency for your technology, start with a Strategy Review.