NIH Wants to Stop Telling You Your Score. Comments on NOT-OD-26-088 Close October 13.

On Friday, August 14, 2026, NIH posted NOT-OD-26-088, a Request for Information on "Proposed Changes to Reporting Outcomes from NIH Peer Review." The proposal is short and its consequence is not. Under it, applicants would no longer receive the final overall impact score or the percentile ranking for their application. In their place, NIH would report one of three categories:

• Most competitive — roughly the top 25 percent for overall impact
• Competitive — roughly the 26th through 50th percentile
• Not discussed — below the 50th percentile, or never brought to discussion

The numerical score and percentile would still be generated inside the review process and retained. They simply would not be released. Reporting on the notice indicates the withholding extends beyond applicants: per Science's account of the proposal, the score would also be unavailable to Institute and Center program staff and to the advisory councils that conduct second-level review.

Comments are due October 13, 2026.

Most of the coverage has framed this as a transparency fight, and it is one. But the operational story for anyone who actually submits to NIH is narrower and more immediate: the number you have been using to decide whether to resubmit is going away, and the number you would have replaced it with — the payline — went away eight months ago.

What you keep, and it is more than the headlines suggest

Read the notice carefully and the loss is specific rather than total. Under the proposal, applicants would continue to receive:

• The written critiques from each assigned reviewer
• The criterion and factor scores from those assigned reviewers
• The summary statement, including the resume of discussion for applications that were discussed

What disappears is the single averaged overall impact score across all voting panel members, and the percentile derived from it.

That distinction matters because the overall impact score was never a mean of the criterion scores. It is an independent holistic judgment, and reviewers are explicitly instructed that it need not track the individual criterion ratings. So for the purposes of revising the science — figuring out what the panel found unconvincing about your approach, your preliminary data, or your investigator team — the criterion scores and the critiques were always doing the heavy lifting. Those survive.

What does not survive is your ability to locate yourself in the field.

The signal you actually lose is competitive position

An overall impact score of 32 with a 19th percentile told you something the critiques could not: how your application compared to everyone else in that round, in a currency that transferred across study sections and across years. It supported four decisions:

1. Resubmit or abandon. A 42nd percentile A0 with a fixable methods critique is a resubmission. A 42nd percentile A0 with a "concept is not compelling" critique is usually a new aim.
2. How hard to revise. Moving from the 30th to the 12th percentile is a different project than moving from the 18th to the 9th.
3. Where to send it next. A percentile that stayed flat across two study sections says something about fit that neither summary statement will say directly.
4. Institutional planning. Research offices use score distributions to forecast, to decide bridge funding, and to counsel junior faculty on tenure timelines.

Now consider how coarse the replacement is. "Most competitive" spans the 1st through 25th percentile. Under the payline regime that band contained both applications that were nearly certain to be funded and applications that were nearly certain not to be. A category that cannot distinguish a 4th percentile from a 24th percentile is not a funding signal. It is a participation ribbon with three colors.

This is the second removal, not the first

The proposal does not land in a neutral environment. On November 21, 2025, NIH issued its Leveraging Funding Policies Framework, effective January 2026, implementing what the agency calls the Unified Funding Strategy. Under it, Institutes and Centers stopped publishing paylines. Funding decisions were explicitly detached from numerical cutoffs and rebalanced toward programmatic considerations: IC strategic plan fit, portfolio balance, public health priorities, applicant career stage.

The practical result, in place since January, is that a strong percentile no longer predicts an award. Applications in the top few percent have gone unfunded; applications well outside any historical payline have been funded. NIH's own IC director commentary in February framed this as flexibility.

Take those two changes together and the sequence becomes clear. In January, applicants lost the threshold — the published number that told you what score was good enough. In August, NIH proposed removing the measurement — the score itself. What remains is a categorical label and a program officer's discretion.

You do not have to believe anything about motive to see the accountability consequence. With scores and paylines both public, anyone could compute the gap between what peer review said and what the agency funded. That gap was auditable in aggregate — by universities, by scientific societies, by journalists, by Congress. Remove the score from the applicant's hands and from the council's, and the gap becomes uncomputable from outside. Whatever the intent, the effect is that programmatic discretion stops being measurable.

That is why this RFI has drawn responses out of proportion to its length, and why it reads differently than it would have in 2019. It arrives after the August 2025 executive order placing senior political appointees in the grant approval chain, and after OMB's May 29, 2026 proposed rewrite of 2 CFR Part 200, which would formally demote peer review to advisory input subject to political pre-issuance review. Each piece is defensible in isolation. The stack is the thing.

What to do in the next eight weeks

Comment, and comment specifically. The RFI is open through October 13. Generic objections to opacity will be counted and set aside. Specific, implementable alternatives are harder to dismiss. Three worth making:

• Narrow the bands. Deciles would preserve most of the claimed flexibility while restoring a usable resubmission signal. Three bands is a design choice, not a requirement of the Unified Funding Strategy.
• Release the score on request after the funding decision. Withholding a number during the decision window is a different policy than destroying the applicant's access to it permanently. A post-decision disclosure preserves program discretion and restores the revision signal.
• Publish aggregate distributions by IC and mechanism annually. Individual scores can be withheld while the agency still reports what percentage of funded awards came from each band. Without this, no one outside NIH can evaluate whether the Unified Funding Strategy is working.

Change your resubmission decision rule now. Do not wait for the policy to finalize — paylines are already gone, so score-based triage is already unreliable. Rebuild the rule around the critiques:

• Count how many of the assigned reviewers raised a concern about significance or innovation versus approach or feasibility. Approach concerns are fixable in an A1. Significance concerns usually are not, and a resubmission that only tightens methods will score the same.
• Read the resume of discussion for whether the panel converged or split. A split panel with one strong advocate is a better resubmission candidate than a unanimous mid-range consensus, regardless of what the impact score was.
• Weight the criterion scores, which you still get, more heavily than you used to. Under the current framework they are the most granular quantitative feedback that will remain.

Move the program officer conversation earlier. Under the Unified Funding Strategy, the PO's judgment already carries more weight than the score. If this proposal is adopted as described, POs may not have the score either — which changes what that conversation is for. Stop asking "was I close?" and start asking "does this fit what the Institute is trying to build?" That question was always the better one; now it is the only one with an answer.

Institutional offices should start their own benchmarking. If NIH stops publishing the number, the only remaining longitudinal record of how your faculty perform in review is the one you keep yourself. Capture criterion scores, band assignments, study section, and outcome for every submission starting with this cycle. In two years that dataset is the only thing that will let you tell a junior PI whether an A1 is worth eleven months.

The honest read

There is a coherent argument for the change. Numerical scores create false precision — the difference between a 22nd and a 26th percentile is well inside the noise of a panel of three assigned reviewers, and treating that difference as meaningful has produced real distortions, including gaming toward reviewer preferences and demoralizing near-miss cycles. Other agencies, including NSF and DOE, have long declined to report percentile-style rankings to applicants. NIH is not proposing something unheard of.

The problem is scale and sequence. NIH runs the largest formalized peer review infrastructure in the world, and its applicant community built its entire resubmission economy around the number. Removing it in the same year the agency stopped publishing paylines, and in the same year OMB proposed making peer review formally advisory, does not read as calibration. It reads as the third step in the same direction.

Comment by October 13. And in the meantime, write the application as though the only thing you will get back is the critiques — because starting now, that is very close to true.