Resubmit, Rewrite, or Walk: How to Read Your NIH Summary Statement
Sometime this month, a founder is going to get a jump scare.
She submitted her SBIR in September, like thousands of other founders. One morning she’ll open her laptop before the coffee is ready and find out what three strangers think of the last year of her company’s life.
For most of the past twenty years, that moment came with a number and a payline. Nobody likes reducing their life’s work to a score. But at least you knew roughly where you stood.
Since January, neither the number nor the payline means what it used to. The summary statements arriving this cycle also look different. Here’s how to read yours.
Where this cycle stands
If you submitted in September, here’s your calendar:
November 2026: your application is reviewed.
About 3 business days later: your score posts in eRA Commons.
January 2027: Advisory Council meets.
April 2027: the earliest your award could start.
Expect your summary statement to take longer than the usual 30 days. After last fall’s shutdown, NIH adopted emergency review procedures, and one of them holds summary statements until later in the Council round. Those procedures run through the January 2027 Council, which is yours. Part of the delay is volume. NIH says the January round drew more than 40,000 applications, well above the previous record.
Which tier you’re in
Every application lands in one of three tiers:
Discussed (roughly the top 30 to 35%). The panel talked about it, and you get an overall impact score.
Competitive but not discussed (the middle third). You get three written critiques but no score. You’re still eligible for funding.
Not competitive (the bottom third).
The document itself is also shorter now. A simplified summary statement has four parts:
One sentence on how much the panel agreed
Bullets listing the points that drove your score
The three reviewers’ written critiques
Your overall impact score, if you were discussed
What a score means without a payline
On November 21, 2025, NIH announced that its institutes would stop using paylines to decide what to fund, starting with the January 2026 Council round. About half of them had published one.
Now every institute decides using six written principles. They include scientific merit (judged from the whole peer review, not just the score), fit with the institute’s priorities, the investigator’s career stage, geographic balance, and how much money is available.
“So my score is meaningless now?”
Not quite. NIH’s own institute directors admit it’s now harder to guess your odds from a score alone. But here’s what changed. A payline looked at one number. Now program staff read the score, the critiques, and the institute’s priorities together.
So the written critiques count for more than they used to. When a percentile settled everything, almost nobody read past it. Now the words matter. The score is still important, but it isn’t the whole decision anymore. That helps founders who read the entire summary statement.
Will NIH keep releasing scores at all? Nobody knows yet. An August request for information proposed replacing scores and percentiles with three categories. At this writing, NIH hasn’t announced a decision.
Your SBIR is scored differently from an R01
In 2025 most academic grants moved to NIH’s Simplified Review Framework. Small business applications didn’t.
Your reviewers still score five criteria separately: Significance, Investigator(s), Innovation, Approach and Environment. Each is on a 1 to 9 scale, where 1 is exceptional and 9 is poor. If your academic friend’s R01 summary statement looks different from yours, that’s why.
Your overall impact score is not the average of those five. It’s the reviewers’ overall judgment of the project.
So if your five criterion scores look decent but your impact score is mediocre, it usually means nobody on the panel got excited. That stings, but it’s useful to know.
Read the statement in this order
1. The tier, on page 1. It tells you whether to expect a score at all, and what kind of conversation to have with your program officer.
2. The consensus sentence and the bullets. This is the panel’s own summary of what decided your outcome. If a point made it into the bullets, it mattered.
3. The five criterion scores, as a pattern. Don’t fixate on any single number. Look at where the high and low scores fall:
Weak Significance and Innovation, strong Approach: they didn’t buy why your problem matters or what’s new about your solution. That’s a framing problem.
Strong Significance, weak Approach: they liked the idea but didn’t trust the plan. Of all the bad news you could get, this is the easiest to fix.
4. The three critiques, side by side. Look for overlap before you react to anything. If two or three reviewers name the same weakness, you have to fix it. If only one does, keep it in mind.
5. The tone, last. Some reviewers sound harsh. This is the hardest part to let go but also the most useless thing to fixate on.
Fixable, harder, or fatal
Sort every weakness into one of three buckets and use the below strategy to help you decide your next move.
Fixable problems usually show up under Approach, Investigator(s) or Environment. Missing controls. A timeline that doesn’t fit the budget period. No backup plan if Aim 1 fails. A commercialization plan that reads like a wish list. A gap on the team.
In each case, you know the answer but it didn’t make it onto the page. A paragraph, a figure, a letter of support or a new collaborator fixes it.
Harder problems show up under Significance and Innovation. The panel didn’t think the problem mattered, or didn’t see what was new. A better Approach section won’t fix that. You have to rebuild your argument for why this work matters. Sometimes you’ll realize that argument fits a different project better.
Fatal is rarer than it feels at 6 a.m. It usually means the panel didn’t believe your premise, because your preliminary data don’t support the core idea. Better writing won’t fix that. New data will.
Here’s a quick test for each major weakness. With the same science and the same team, could you have avoided it by writing differently? If yes, it’s fixable.
When one reviewer sinks you
Two reviewers liked it. The third wrote four paragraphs about a control you think is irrelevant, and that reviewer decided your outcome.
Was that just bad luck? Partly, yes.
In a 2018 PNAS study, Elizabeth Pier and colleagues had separate panels review the same NIH applications. Individual reviewers’ scores showed essentially no agreement. (Oof.)
Critics pointed out, fairly, that every application in the study was strong, which makes disagreement more likely. But that’s exactly your situation. Strong applications near the funding line are the ones you’re competing against.
That doesn’t mean the process is random. Better applications still do better on average. It just means any single review is a noisy measurement, so one bad draw doesn’t prove your application deserved it.
What to do about it: in your resubmission, respond to every concern, but only restructure the application around concerns that more than one reviewer raised. Rebuilding a solid application to please one reviewer is the most common way resubmissions get worse.
Your four options
1. Resubmit. NIH allows one resubmission, within 37 months, and only after you have your summary statement. Do it when your weaknesses are fixable and the panel showed real interest. Resubmissions get funded at higher rates than first submissions, partly because the applicants who resubmit tend to be the ones who scored well.
2. Take it to a different funder. Do this when your science is solid but this panel was the wrong audience. Your summary statement still tells you which arguments didn’t land, so you know what to change for the next audience. Our guide to non-SBIR funding [BLOG URL] covers the options.
3. Start fresh with a new application. You can resubmit an unfunded application as new. After an unsuccessful resubmission, NIH will accept the same idea as a new application without checking how similar it is. You give up the Introduction page and the chance to respond to reviewers. This makes sense when most of the damage was under Significance and Innovation.
4. Shelve it. If the panel doubted your premise and, after two weeks away, you partly agree, the review just gave you useful information. Spend the next cycle running the experiment that answers their question.
Write an Introduction that admits what was wrong
Your Introduction has to do three things on one page: summarize the major changes, respond to the reviewers’ issues, and fit on a page. You’re not allowed to mark changes in the body of the application, so this page is the only place reviewers see what changed.
A few rules:
Concede cleanly. “Reviewers noted the Aim 2 timeline was not realistic. We extended it to nine months and removed the second cell line.”
Show the change, not the intention. “We added dose-ranging data in Figure 3” is a change. “We strengthened our dosing approach” is not rigorous enough.
Push back once at most, and only with new evidence.
Take the blame for any confusion, even if you suspect it was theirs. It costs you nothing.
How long this really takes
If your summary statement arrives in December or later, you technically could still make January 5. But a resubmission written over the holidays in three weeks is how applications get worse. For most September applicants, April 5, 2027 is the realistic target.
Three rules affect your plan:
HHS accepts at most nine new and resubmission Phase I, Fast-Track and Direct to Phase II applications per small business per fiscal year. A resubmission takes up one of those slots.
NIH caps each PI at six applications per calendar year.
NIH no longer accepts late small business applications, for any reason.
What to do, in order
The day your statement arrives: read the tier and the bullets. Nothing else.
The next day: read the three critiques side by side.
Before you talk to anyone: sort each weakness into fixable, harder or fatal.
Email your program officer one question: “Where does my application sit relative to the institute’s priorities this year?” Without paylines, your program officer has the information a payline used to give you.
Pick one of the four options with your calendar open in front of you.
Got a summary statement coming and want a second reader there when you open it?
Comments