NIH Impact Score vs. Percentile: What Both Numbers Mean and How to Use Them

How NIH Assigns an Impact Score

The impact score posted to eRA Commons is the average of scores from your assigned reviewers — typically three — multiplied by 10. Each reviewer uses a 9-point scale where 1 is exceptional and 9 is poor. The resulting impact score runs from 10 to 90, with lower always meaning better. A score of 10 essentially never appears in practice. Competitive scores almost always fall between 15 and 55, and scores below 25 represent the genuinely competitive range for most institutes in most years, though that benchmark has shifted since paylines disappeared.

Applications that are triaged — meaning they receive sufficient negative written reviews to be withdrawn from in-person discussion — do not receive a score. You'll see "ND" (Not Discussed) in eRA Commons instead. This is meaningfully different from receiving a poor score. It means at least two of your three reviewers ranked the application in the lower half of the reviewed pool before the meeting, and the panel did not bring it to discussion. Your summary statement will still include written critiques, and those critiques should shape your resubmission planning more than the absence of a number. Not Discussed does not mean the science is bad; it often means scope, framing, or preliminary data did not clear the bar that specific panel needed.

What the Percentile Rank Actually Measures

The percentile is not a normalized version of your impact score. It ranks your score against all scored applications from the same study section across the current review round and the two preceding rounds, which typically spans about 12 to 18 months of data. This rolling three-cycle comparison is the core mechanism that separates percentile from raw score. It exists specifically to counteract "score creep": the tendency for review panels to give progressively better scores to applications of similar quality over time, causing raw scores to cluster in the excellent range and lose the ability to discriminate between applications.

Percentile ranks run from 1 to 99 in whole numbers, with lower meaning better. A 15th percentile means your score ranked better than 85 percent of scored applications your panel reviewed over the past three cycles. One detail that trips researchers up: not every scored application receives a percentile. Review panels that are not standing study sections — some special emphasis panels, certain training grant reviews, and some expedited review mechanisms — may generate only an impact score with no percentile. If your program announcement specified a special review, check the funding opportunity for whether a percentile will be assigned before you factor it into your planning.

Why Your Score and Percentile Can Tell Different Stories

Score inflation is real, and it varies by study section. A panel that has been assigning generous scores for several cycles will produce higher percentile ranks for the same raw score relative to a panel with compressed scoring. Concretely: a raw score of 22 at one study section might land at the 18th percentile. That same raw score at a study section known for tighter scoring might come back at the 32nd. If you're comparing your outcome to your mentor's experience a decade ago, or to a colleague reviewed by a different panel, you need to account for this. The percentile is usually the more informative number when comparing across reviewers or cycles because it corrects for panel-level inflation. The raw score is more informative when comparing two rounds from the same study section in close succession.

There's also a direction-of-travel question. If your study section has been gradually tightening scores over recent cycles, a current score of 25 may carry more competitive weight than a 25 issued two years ago by the same panel — even though the numbers look identical. Your program officer can often tell you whether your study section's scoring patterns have shifted recently. That is a legitimate question to raise in a post-score call, and the answer changes whether a resubmission strategy should aim for targeted revisions or a deeper rethink of framing and approach.

The Simplified Review Framework Changed What Goes Into Your Score

For applications with due dates on or after January 25, 2025, NIH reorganized the five traditional review criteria — Significance, Investigators, Innovation, Approach, and Environment — into three scoring factors: Importance of the Research, Rigor and Feasibility, and Expertise and Resources. Reviewers still evaluate the same underlying science, but the written critique structure and the way criterion scores map to the Overall Impact score changed. If you're reading a recent summary statement alongside an older one from a mentor or colleague, the section headers and internal structure will look different.

What this means strategically: Innovation used to be its own scored criterion, giving reviewers an explicit structural hook to argue for or against novelty. Under the new framework, novelty arguments sit inside "Importance of the Research." Some investigators find this helps because incremental-novelty concerns are now bundled with significance rather than flagged as a standalone weakness. Others find it harder to address clearly in a resubmission because the critique is less compartmentalized. Either way, read your critiques with the new factor structure in mind before you draft the A1 introduction. A reviewer who flagged weak innovation under the old framework may have expressed the same concern very differently in 2025 or 2026 language, and missing that mapping is an easy way to produce an introduction that addresses the wrong thing.

How Funding Decisions Actually Work Without Paylines

Before early 2026, most NIH institutes published an explicit payline: a percentile threshold below which every clean application in the normal funding cycle was awarded. Paylines varied by institute and year, but they gave applicants a reasonably predictable answer to the question "am I going to get funded?" The Unified Funding Strategy changed this. Institute and Center Officials (ICOs) now weigh scientific merit alongside career stage, geographic diversity, the institute's existing portfolio balance, and specific scientific priorities. No single percentile threshold determines the outcome.

In practice, this means an application at the 28th percentile might be funded at an institute whose portfolio has an underrepresented area matching your work, while a 20th percentile application in a well-covered area might not. The percentile still matters — it is still the primary input to merit evaluation — but it is no longer a deterministic cutoff. Applications in what used to be the gray zone just above the old payline now carry more uncertainty than before, and applications that would have been automatic awards under the old system now require the same program officer conversation that borderline cases always did.

There is still a mechanism called "select pay" — applications outside the normal competitive range that an institute funds because they address a specific scientific priority. This existed before the Unified Funding Strategy and continues under it. If your application addresses a topic your target institute has publicly flagged as a current priority, your program officer can advocate for select pay consideration even at a score that wouldn't ordinarily reach funding. That conversation is worth having directly and early, not as a last resort after waiting for the award notice.

Reading Your Numbers for the Resubmission Decision

The score is not the resubmission decision — the written critiques are. But the score and percentile together tell you how much ground you need to make up, which shapes the scale of revision worth writing. A rough framework, with the caveat that every situation differs and a program officer call is irreplaceable:

One more thing worth knowing: the percentile itself can shift slightly after each review cycle, because the rolling three-cycle comparison updates when the study section meets again. Your impact score will not change, but the percentile assigned to it may tick up or down by a point or two. If you're watching an application you hope gets funded on a second look, small percentile fluctuations are a normal artifact of the calculation and almost never signal anything actionable on their own.

Frequently Asked Questions