NSF GRFP Formatting: What Gets Returned Unreviewed

NSF GRFP Formatting Rules That Get an Application Returned Without Review

Most fellowship advice is about what to write. This is not that. The NSF Graduate Research Fellowship Program publishes a short list of mechanical requirements whose failure mode is not a lower score — it is the phrase "returned without review," which the current solicitation prints in at least five separate places. An application can be strong, finished, uploaded on time, and still never reach a panelist because two headings were combined into one.

The rule that catches the most people is the heading rule. NSF requires separate "Intellectual Merit" and "Broader Impacts" sections, under individual headings, separated by a line break — in both statements, not just the research plan. Applications that do not conform are returned without review.

Below is every mechanical rule stated as the sponsor states it, what the application module checks automatically, where the official templates live, and a verification pass to run on the two PDFs before you upload them. This is a proofreading procedure, not writing advice.

Verification banner. Every rule on this page was verified against NSF 25-547, the GRFP program solicitation posted September 26, 2025, which covers the FY2026 competition. Retrieved 2026-08-19. As of that date NSF has not posted an FY2027 GRFP solicitation — no NSF 26-xxx GRFP solicitation exists yet, so no next-cycle deadline exists to quote. Formatting rules are solicitation-versioned and have measurably changed between cycles (see what changed). Re-verify every number and every rule below against the current solicitation before you submit anything. Where this page shows dates, they are the dates printed in NSF 25-547 for the FY2026 competition — they are not "the 2027 deadlines," and no such deadlines have been published.

Table of Contents

• The rule that returns applications
• What "individual headings separated by a line break" means
• The page limits, and what counts against them
• The rest of the format contract
• The module checks first, and that is not a safety net
• The official templates, and where they live
• What changed between the last two solicitations
• Two deadlines, and the letter one comes first
• Non-formatting rules that also stop a review
• The pre-upload verification pass
• What a review can and cannot check
• Frequently Asked Questions

The rule that returns applications

NSF 25-547 states it inside the list of statement formatting guidelines:

"Separate sections for 'Intellectual Merit' and 'Broader Impacts,' under individual headings and separated by a line break must be included in both the Personal and Research Plan statements. DO NOT include other terms in the individual headings, combine these two sections into one section, or combine them with any other section"

The capitalization is NSF's. Two sentences later, the same list gives the consequence twice:

"Personal and Research Plan statements must have line breaks between sections. Applications that do not have line breaks between sections will be returned without review; and"

"Applications that do not conform to the heading and section requirements for Intellectual Merit and Broader Impacts will be returned without review."

Read the scope carefully, because it is the part that gets missed. The requirement is not "address both criteria somewhere." It is four discrete demands:

1. Separate sections — one for each criterion, not a blended paragraph.
2. Individual headings — a visible heading for each, not a topic sentence.
3. Separated by a line break — a structural break, not just a new paragraph of prose.
4. In both statements — the three-page Personal, Relevant Background and Future Goals Statement and the two-page Graduate Research Plan Statement.

The "DO NOT" clause adds a fourth failure most drafts walk into voluntarily: the headings must be those terms and not decorated versions of them. A heading reading "Intellectual Merit and Broader Impacts of the Proposed Work" violates the rule three ways at once — it adds other terms, it combines the two sections, and it leaves no line break between sections that no longer exist.

For what each section is supposed to contain, NSF's merit review criteria are one sentence each: "The Intellectual Merit criterion encompasses the potential to advance knowledge," and "The Broader Impacts criterion encompasses the potential to benefit society and contribute to the achievement of specific, desired societal outcomes." That substance question is a different job from this one — what each section has to prove, and what counts as evidence for each, is worked through in Intellectual Merit vs Broader Impacts, and the same split, seen from the recommender's chair, in the all-merit, no-Broader-Impacts letter.

What "individual headings separated by a line break" means

Here is the honest boundary: NSF publishes the requirement, not a word-processor recipe. The solicitation does not specify a heading style, a font weight, a numbering scheme, or a mechanism for producing the line break. What it specifies is that a break exists between sections and that each criterion sits under its own heading bearing only that term.

So the safe reading is the literal one — a visible break and a bare heading — and the arbiter is the module's automated check, not any interpretation offered here or anywhere else.

Constructed illustration — invented for this guide, not an excerpt from any real application.

Conforming shape:

…final paragraph of the preceding section.
 
Intellectual Merit
 
Text of the Intellectual Merit section.
 
Broader Impacts
 
Text of the Broader Impacts section.

Shapes that do not conform to the quoted rule:

• Intellectual Merit and Broader Impacts — one heading, two sections combined.
• Intellectual Merit of My Proposed Research — "other terms in the individual headings."
• Broader Impacts & Future Goals — combined with another section.
• A bolded run-in lead like **Intellectual merit.** This project will… on the same line as the body text — no line break separating the section.
• Both criteria present and well argued, but only in the research plan — the requirement says both statements.

Two practical notes that follow from the rest of the format list rather than from the heading rule itself. The margin rule bans text inside the 1" margins including headers and footers, so a heading placed in a running header is a second violation, not a clever solution. And because the module rejects non-compliant documents outright, the time to test a heading structure is a week before the deadline, not on deadline night.

The page limits, and what counts against them

The sentence people skip is the one about what fills those pages:

"These page limits include all references, citations, charts, figures, images, and lists of publications and presentations."

There is no bonus citations page. This is the sharpest structural contrast with the other big fall STEM fellowship: the NDSEG research proposal is a three-page body plus a separate fourth page for cited work, and it is uploaded twice — once original, once redacted. GRFP has no redaction requirement at all, and no free page. A reference list, a figure, and a publications list all eat the same three or two pages your argument needs.

Two consequences worth planning around:

• A figure is a budget decision. A half-page figure in a two-page research plan is a quarter of the document. It has to earn that against text.
• Citations compete with content. NSF requires that "Cited references must include the name of the journal or publication (abbreviations accepted)," and that "URLs and DOI references will not be accepted in place of the name of the journal" — so you cannot compress a reference list into bare DOIs to save space.

If cutting is where you are stuck, the meaning-first approach in how to shorten a statement without losing its evidence transfers directly: cut restated credentials and abstraction, not the specific decisions and outcomes that are the only thing a panel can actually evaluate.

The rest of the format contract

Quoted from the statement formatting guidelines in NSF 25-547, in the solicitation's own wording:

And the catch-all, in the same section:

"Applications that are not compliant with these format requirements will be returned without review."

Four items on that list are the ones that surprise people:

No text inside the margins, including page numbers. A word processor's default header or footer, a name block, or automatic pagination all place text inside the 1" margin. This is the single most common accidental violation, because most people never inserted the footer deliberately — a template or a document style did.

No URLs or weblinks. Not in the body, not as a portfolio pointer, not as a link to your lab site or your preprint. The rule extends past the link itself to "direct the reviewer to external sites."

Six lines per inch, maximum. This is a density ceiling that operates independently of the font-size floor. Setting 11-point type on tight leading to gain a paragraph fails this rule even though the font size is legal.

A narrow font list. Arial Narrow is excluded by name. If your draft is in Calibri, Helvetica, Georgia, Garamond, or your journal template's house font, it is not on the list, regardless of point size.

The module checks first, and that is not a safety net

NSF 25-547 says the checking is automatic:

"Compliance with these guidelines will be automatically checked by the GRFP Application Module. Documents that are not compliant will not be accepted by the GRFP Application Module."

That sounds protective. Read the next sentence in the solicitation and it stops sounding that way:

"Non-compliant documents will block upload of the complete application for receipt by the deadline. Applications that are not compliant with these format requirements will be returned without review."

The automated check is a gate on upload, not a proofreader you can lean on. If the module refuses your PDF at 4:40 p.m. on your field's deadline day, the check did not save you — it converted a formatting problem into a missed deadline. The practical implication is a scheduling one: attempt a real upload days early, so that a rejected document is a Tuesday problem instead of a terminal one.

Applications are submitted "online using the NSF Graduate Research Fellowship Program Application Module at research.gov/grfp," and NSF adds that "Only the information required in the GRFP Application Module will be reviewed."

The official templates, and where they live

NSF publishes Word templates for both statements. Use the sponsor's files rather than any reproduction of them, including this page's — a template is a versioned artifact, and a copy made by a third party is a snapshot of whichever cycle it was copied from.

• GRFP Personal Statement Template (.dotx)
• GRFP Graduate Research Plan Template (.dotx)
• Both are listed on the program's Applicant Additional Resources page, alongside the Statements page, and the same Personal Statement template is mirrored on research.gov.

Three cautions about templates generally, official ones included:

1. A template is a starting point, not a compliance certificate. Nothing in the solicitation says a document built from the template is automatically compliant, and the module still checks the file you upload. Editing inside a template can reintroduce a footer, a font substitution, or a heading change.
2. Font substitution is silent. Opening a .dotx in a different word processor, or on a machine missing Palatino Linotype or Computer Modern, can swap in a font that is not on NSF's list without warning you.
3. Verify the export, not the editor. The rules apply to the PDF. Check the PDF.

Third-party templates deserve more suspicion, not less, for a reason the next section makes concrete.

What changed between the last two solicitations

This is why the banner at the top of this page exists. The formatting rules are not stable across cycles, and the previous solicitation is still indexed, downloadable, and quoted by university advising pages. Comparing NSF 24-591 with the current NSF 25-547:

Every one of those deltas is a way for a page written one cycle ago to be confidently, specifically wrong. A guide that tells you to set Times New Roman because "that is the required font" is describing NSF 24-591. A guide that shows mid-October deadlines is describing a competition that closed in 2024.

The operating rule: check every mechanical number against the solicitation currently linked from NSF's GRFP program page, including the numbers on this page. Where the cycle currently stands, and what is safe to draft before a new solicitation posts, is tracked in the GRFP cycle guide.

Two deadlines, and the letter one comes first

This is the scheduling fact that most surprises applicants, and it looks like a typo until you check a calendar.

In NSF 25-547, the application deadlines are staggered by field and run from November 10 through November 14, 2025, "received by 5 p.m. local time of applicant's mailing address":

The reference letter deadline is a separate, earlier date:

"Reference letters are due Friday, November 7 at 5:00 p.m. Eastern Time (ET); 4:00 p.m. Central Time; 3:00 p.m. Mountain Time; 2:00 p.m. Pacific Time; 12:00 noon Hawaii Time; or time corresponding to 5:00 p.m. Eastern Time in all other time zones."

That sentence omits its year, which is what makes it read as an error. It is not one: November 7, 2025 was a Friday, and it falls three to seven days before every application deadline in the table above. In this solicitation, letters were due before the application itself — and the letter deadline was a single fixed date for everyone, while the application deadline varied by field.

The consequence of missing it is not a formatting matter but has the same effect: "Two reference letters from non-family members must be received by the reference letter deadline for an application to be reviewed. If fewer than two reference letters (one or none) are received by the reference letter deadline, the application will not be reviewed."

Cycle-hedge, stated plainly: the dates above are the ones printed in NSF 25-547 for the FY2026 competition. They are not next cycle's dates, and NSF has published no next-cycle dates as of 2026-08-19. What is durable is the shape — a fixed, earlier letter deadline and later field-staggered application deadlines — and even that must be re-checked when a new solicitation posts.

The rest of the letter mechanics — the two-page cap the module enforces, the font and letterhead rules, how many writers to list, and why the applicant's ranking of writers decides which letters get read — are covered in the GRFP reference letters guide rather than repeated here.

Non-formatting rules that also stop a review

Formatting is not the only route to "returned without review." The same solicitation lists several others, and they belong on the same pre-submission checklist because the outcome is identical:

• Eligibility. "Applications that do not meet eligibility requirements will be returned without review as being ineligible for a fellowship."
• Field-of-study alignment. "Applications submitted to a Field of Study deadline not in alignment with the proposed research plan and degree program will be returned without review." Because the deadlines are staggered by field, choosing a field to buy four extra days is choosing a rejection.
• Completeness and status. "Applications that are incomplete due to missing required transcripts and/or reference letters (fewer than two letters received), or that do not have 'received' status in the Application Module on the application deadline for the selected Field of Study will not be reviewed." Received status is the bar — not saved, not uploaded, not in progress.
• Who reads it. "Applications will be reviewed by experts in the first Major Field of Study and associated subfield listed." The field selection is a routing decision as well as a deadline.

The pre-upload verification pass

Run this on the exported PDFs, not on the documents in your editor. It takes about fifteen minutes per statement and it is the entire point of this page.

Both statements, one at a time:

1. Headings. Search the PDF for Intellectual Merit and Broader Impacts. Confirm each appears as its own heading, in that wording, with no additional words in the heading, in this document. Repeat for the other statement.
2. Line break. Confirm each of those headings is separated from the preceding section by a break, and that neither heading runs inline with body text.
3. Not combined. Confirm the two are not merged with each other or folded into another section.
4. Page count. Three pages maximum for the Personal statement, two for the Graduate Research Plan — counting references, citations, charts, figures, images, and any publications or presentations list.
5. Margins. Confirm 1" on all four sides and no text inside them. Look specifically for an auto-inserted page number, a header, a footer, or your name in a running head.
6. Font. Check the PDF's font properties, not your editor's font menu, and confirm what actually embedded is on NSF's list at or above its size floor. Confirm Arial Narrow did not substitute for Arial.
7. Density. Confirm no more than six lines of text within a vertical inch, and that no line-spacing setting is below single.
8. Links. Search the PDF for http, www, doi, and .com. Remove URLs and weblinks, and confirm cited references name the journal or publication rather than pointing at a DOI.
9. Images. Confirm every image, graph, and chart carries descriptive alternative text and adequate color contrast.
10. File type. PDF only.
11. Upload early. Attempt the real upload well before your field's deadline, because a non-compliant document blocks the upload rather than warning you.

Application-level, same session:

12. Confirm the field of study you selected aligns with the research plan and degree program, and note which deadline that field carries.
13. Confirm transcripts are attached and the application shows received status.
14. Confirm at least two reference letters are in, against the letter deadline — which in NSF 25-547 fell before every application deadline.

That is a checklist you run. It is not a guarantee about any tool, any template, or any outcome.

What a review can and cannot check

Be precise about the boundary, because this page is about mechanics and a text review is not a mechanical check.

A rubric review reads the text you paste in. It cannot open your PDF, measure margins, detect an embedded font, count pages, see a footer, verify alternative text, or confirm that the application module accepted anything. Nothing here substitutes for opening the current solicitation and the exported file yourself, and nothing here substitutes for your advisor's read of the science.

What a review is useful for is the layer above the mechanics: whether the Intellectual Merit section actually argues potential to advance knowledge rather than listing techniques, whether the Broader Impacts section names a specific desired outcome and evidences an action you took rather than promising future outreach, and whether the two sections say different things instead of paraphrasing each other. The GRFP personal statement rubric and the GRFP research plan rubric read for exactly that, on the draft you supply and nothing else. Feedback is a revision signal, not a verdict on your application, and no review changes what a panel decides.

More STEM fellowship guides

• NSF GRFP cycle guide — where the cycle stands and what is safe to draft before a solicitation posts
• NSF GRFP reference letters: what reviewers score — the two-page cap, the counting rules, and the separate letter deadline
• The all-merit, no-Broader-Impacts letter — the same two-criterion split, from the recommender's chair
• NDSEG research proposal: format and redaction — the contrasting mechanical regime: four pages, dual upload, and a redaction rule GRFP does not have
• NDSEG fellowship essays guide — two typed 500-word boxes and the cycle window
• GRFP, NDSEG, Hertz in one season — the deadline ladder and what evidence transfers
• How to write a scholarship essay that fits the award
• How a scholarship review works

Review your draft

Once both statements are conforming and exported, the writing is the remaining variable. Paste a statement into the scholarship essay review and run it against the GRFP personal statement rubric or the GRFP research plan rubric; both documents can be reviewed through the NSF GRFP application review. Two Quick Reviews are free each day; your first Full Review is $5, once per account, and Pro is $50 for ten Full Reviews. Reviews are of your own writing only, and feedback is a revision signal rather than a verdict on your application.

Frequently Asked Questions

Do the Intellectual Merit and Broader Impacts headings go in both statements?

Yes. NSF 25-547 requires separate sections for "Intellectual Merit" and "Broader Impacts," "under individual headings and separated by a line break," and says they "must be included in both the Personal and Research Plan statements." Applications that do not conform to the heading and section requirements "will be returned without review."

Can I title the heading "Intellectual Merit of My Research"?

No. The solicitation says "DO NOT include other terms in the individual headings, combine these two sections into one section, or combine them with any other section." Use the bare terms.

How long are the two GRFP statements?

Under NSF 25-547, the Personal, Relevant Background, and Future Goals Statement is "three (3) pages (PDF)" maximum and the Graduate Research Plan Statement is "two (2) pages (PDF)" maximum. Those limits "include all references, citations, charts, figures, images, and lists of publications and presentations" — there is no extra citations page.

Is there an official NSF template?

Yes — a Personal Statement template and a Graduate Research Plan template, both .dotx files published on nsfgrfp.org and mirrored on research.gov. Download them from the sponsor rather than from a third-party copy, and still verify the exported PDF: the module checks the file you upload, not the template you started from.

Will the application module tell me if my formatting is wrong?

It checks, but not as a courtesy proofread. NSF 25-547 says "Compliance with these guidelines will be automatically checked by the GRFP Application Module. Documents that are not compliant will not be accepted by the GRFP Application Module," and that "Non-compliant documents will block upload of the complete application for receipt by the deadline." A rejection on deadline day is a missed deadline, so upload early.

Can I include a link to my lab page or a preprint?

No. NSF 25-547 says "Do not include URLs or weblinks or direct the reviewer to external sites." Cited references "must include the name of the journal or publication (abbreviations accepted)," and "URLs and DOI references will not be accepted in place of the name of the journal."

Are the reference letters really due before the application?

In NSF 25-547 they were. The solicitation states "Reference letters are due Friday, November 7 at 5:00 p.m. Eastern Time (ET)," while the application deadlines in the same document ran November 10–14, 2025 by field — so the letter deadline preceded every application deadline by three to seven days. That is solicitation-specific: re-check both deadlines in the current solicitation, since no next-cycle dates have been published as of 2026-08-19.

What are next cycle's GRFP deadlines?

There are none to quote yet. As of 2026-08-19, NSF 25-547 — posted September 26, 2025 for the FY2026 competition — is still the solicitation shown on NSF's GRFP program page, and no FY2027 GRFP solicitation has been posted. Any page giving you a specific next-cycle date is either citing something we could not find or inventing it. Watch NSF's GRFP program page for the new solicitation, and treat every rule on this page as provisional until you have checked it there.

Sources

All retrieved 2026-08-19.

• NSF — GRFP Program Solicitation NSF 25-547 (posted September 26, 2025; FY2026 competition) — statement formatting guidelines, heading and line-break requirement, "returned without review" clauses, page limits, application and reference letter deadlines, module compliance checking, merit review criteria
• NSF — GRFP program page — current solicitation status; no FY2027 solicitation posted as of retrieval
• NSF — GRFP Program Solicitation NSF 24-591 — prior-cycle formatting rules, used only for the cycle-to-cycle comparison
• NSF GRFP — Statements and Applicant Additional Resources — page limits, heading requirement, official template listings
• NSF GRFP — Personal Statement template, Graduate Research Plan template, and the research.gov mirror
• NSF — GRFP Application Module — submission platform named in the solicitation