References in a grant application are governed by the funder, not by a style manual, and funders disagree on nearly every point: whether the list counts toward the limit, whether links are allowed, and which identifiers it must carry. This guide, published by EdCitation, sets out the rules of the NIH, NSF, UKRI, the ERC and the Australian Research Council, each read at the funder's own pages on 27 September 2026, and says what could not be confirmed.
It is the reference-list companion to how to write a grant proposal, which covers sections, limits and review criteria. When the list is finished, EdCitation's free Verify references looks every entry up in the publisher's record and flags retracted papers, with no account. EdCitation looks references up and never writes any part of an application: every sentence a reviewer reads stays yours.
Do references count toward the page limit in a grant application?
At the NIH, NSF and ERC they do not; in UKRI's Funding Service they count inside each question's word limit unless the call gives them a question of their own. The table sets out each rule as the funder's pages state it.
| Funder | Where references go | Style set? | Count toward the limit? | Special requirement | Source |
|---|---|---|---|---|---|
| NIH | Bibliography & References Cited attachment | None named | No: no page limit | PMCID or NIHMSID for your own NIH-funded papers; no active hyperlinks | G.220 form (FORMS-I); page limits table |
| NSF | References Cited section | None; the elements are named | No: no page limit | Every author in published order; citations only | PAPPG 24-1, Chapter II |
| UKRI Funding Service (BBSRC call read) | Inside the question they support | "As appropriate to your research" | Yes, in that question's word count | Persistent identifiers where possible; links only to reference information | UKRI guidance; BBSRC call |
| ESRC (a UKRI council) | A References question of its own | None named | Own limit: 1,000 words | Every reference there and nothing else; links advised against | ESRC research grants call |
| ERC Starting and Consolidator | Inside Part B1 and Part B2 | One common in your discipline | No | Same PDF as the proposal | ERC form; Information for Applicants |
| ARC | Body and Research Outputs list | Not confirmed | Not confirmed | Preprints identified, with DOI or URL and version or date | ARC communiqué, 2021 |
| The call you answer | Check your paper lists its limits, free | Verify references checks each entry |
The call always wins. The National Institutes of Health (2024c) says a notice of funding opportunity overrides its application guide, and that an attachment has no limit only where none is listed there, in the notice or in a related notice.
NIH: the Bibliography & References Cited attachment
For a research grant, NIH asks for one PDF holding every reference cited in the Research Plan and the human subjects form, with no page limit. We read the rule in form G.220 of the FORMS-I instructions, in an archived copy of 24 January 2025, because NIH's grants site answered our reader with a security check we did not try to pass; any later change to the wording could not be confirmed.
No page limit, but no padding and no live links
The National Institutes of Health (2024b) sets no page limitation but asks you to stay concise and cite relevant, current literature pertinent to the work. Its two lines on links sit uneasily together: the section opens by barring URLs and hyperlinks unless the notice allows them, then lets a work that is free online carry a URL or PubMed ID (PMID) and bars only active hyperlinks. Our reading, not NIH's: plain-text identifiers, no clickable links, unless your notice says otherwise.
The PMCID rule under the NIH Public Access Policy
For papers you wrote or co-wrote that arose from NIH support and fall under the Public Access Policy, each reference carries its PubMed Central number (PMCID) or its NIH Manuscript Submission number (NIHMSID). Where the journal deposits in PMC for its authors and no PMCID exists yet, the form asks for "PMC Journal - In Process".
The policy changed in 2025. Under the 2024 NIH Public Access Policy, the accepted manuscript goes to PubMed Central on acceptance, public from the official date of publication with no embargo, and an NIHMSID may stand in until the PMCID is assigned (National Institutes of Health, 2024a). The start date moved to 1 July 2025 for manuscripts accepted from then on, according to the National Institutes of Health (2025). Noncompliance may weigh in future funding decisions and can delay continuation awards.
NCBI's PMC ID Converter turns a DOI, PMID or NIHMSID into the others. For this guide it returned PMC5866547 for the DOI of Pier et al. (2018), a study of NIH review, and no PMCID for a 2013 Nature Reviews Neuroscience paper that is not in PMC.
Preprints and other interim research products
NIH lets you cite a preprint wherever publications are cited, this attachment included. The National Institutes of Health (2017) asks for the DOI and the object type, such as "Preprint", plus version details and, where relevant, the date cited. Preprints fall outside the Public Access Policy, so they need no PMCID.
NSF: the References Cited section in the PAPPG
NSF makes References Cited a required section, and a proposal missing a required section is not accepted or is returned without review. The rules below are from PAPPG 24-1, which NSF's page still calls current, though draft guidance to replace it was open for comment until 24 August 2026 (National Science Foundation, n.d.).
Every author, in the order published
The National Science Foundation (2024) requires every author, in the order of the publication, with the article and journal title, book title, volume, pages and year. That rules out "et al." in an NSF list, whatever your style usually does; when to use et al. covers the styles. A web address is optional.
Citations only, and no page limit
There is no set page limit, but the section may hold bibliographic citations only: notes moved there to escape the 15-page Project Description break the rules. Publications from an earlier NSF award may be cited in full either under Results from Prior NSF Support or in References Cited.
UKRI: references inside the word count
In UKRI's Funding Service references sit inside the answer they support and count toward its word limit. UK Research and Innovation (2026) asks you to prioritise the most pertinent, make each easy for assessors to identify, format them as your field does and use persistent identifiers where possible. Its examples show both an in-sentence reference and a footnote, so no style is imposed.
BBSRC's standard research grant call shows the cost. Vision has 550 words and Approach 3,300, each of which may include references, and the R4RI modules 1,150 words, references included (Biotechnology and Biological Sciences Research Council, 2026).
When a council gives references their own question
ESRC does the opposite. Its research grants call has a References question of up to 1,000 words, wants every reference there and none in the other answers, and allows nothing else in it (Economic and Social Research Council, 2026). One funder, two rules.
Links and persistent identifiers
UKRI allows hyperlinks only to reference information, and assessors need not open them. ESRC advises against links altogether and suggests a DOI where you do link. A DOI written as an identifier satisfies both; what a DOI is explains why it outlasts a web address.
ERC: references in Part B1 and Part B2
The ERC keeps references outside every page limit and inside the same document. The European Research Council (2025) caps Part I of the scientific proposal (Part B1) at five pages and Part II (Part B2) at seven, references not counted, and asks for a style common in your discipline that lets evaluators retrieve each work, naming ACS, AMA and MLA as examples.
Do not split them off
The form warns against uploading references separately: at Step 1 reviewers receive one document, and a separate attachment never reaches them. The Information for Applicants for the 2027 calls, version 11.0 of 22 July 2026, repeats that references are outside the limits and says reviewers need not follow links to web pages (European Research Council, 2026). The CV's track record lists up to ten research outputs, preprints, datasets and software included.
What happened when the ARC ruled out preprints?
In August 2021 the Australian Research Council declared 17 of 675 Future Fellowships applications and 15 of 996 DECRA applications ineligible for citing preprints, which its rules then barred, as the Australian Research Council (2021b) itself reported. On 14 September 2021 it reversed course for future rounds: preprints may be cited anywhere, the Research Outputs list included, and assessors judge their worth (Australian Research Council, 2021a).
Each must be identified as a preprint and listed with a DOI, URL or equivalent and a version number or access date, as applicable. We read both statements in archived copies because the ARC's site did not respond on 27 September 2026, and its current page limits, in Instructions to Applicants on GrantConnect, were not read. Types of academic sources explains what a preprint is.
What does it mean when a funder sets no citation style?
It means the choice is yours and consistency is the rule. NIH names no style, NSF names elements, UKRI asks for your field's format and the ERC gives examples. So use your discipline's style (a biomedical application often uses Vancouver, which Cite a source in Vancouver builds from a DOI or PubMed ID), add what the funder adds (NSF's full author list, NIH's PMCIDs, ARC's preprint versions), keep one style throughout, and make every work findable. The same holds for a doctoral proposal, where the university sets the rules (PhD applications), and for a journal's list (the manuscript checklist).
How to check a grant application's references before submission
Check the list last, after the narrative is frozen.
- Read the call's reference rule. For the NIH, read the notice of funding opportunity. EdCitation's free Check your paper reads a call into a checklist of limits and sections, each beside its sentence.
- Build each entry from its record. Cite a source makes it from a DOI, PubMed ID, ISBN or title, in APA 7, MLA 9, Chicago 18 author-date, Harvard, IEEE or Vancouver.
- Restore every author for NSF.
- Add PMCIDs to your own NIH-funded papers, finding each in the PMC ID Converter; Verify references then checks it.
- Mark preprints with DOI, type, version and date.
- Remove live links where barred, keeping identifiers as plain text.
- Look every entry up in Verify references, or upload the application so each citation in the narrative is matched to the list and back.
- Count again for UKRI, where references use your words.
How EdCitation checks a grant application's reference list
EdCitation is, in our view, the best tool for this job, because it looks each reference up in the publisher's record and never writes one, which no chatbot can do; and NSF's demand for every author in published order is exactly what a record shows. Verify references is free with no account: each entry comes back verified, doubtful (shown as "check this") or not found, retracted papers are flagged, and one it could not check is never shown as not found. References from a file, part of Pro at $8 a month with three checks a month on a free account, checks a whole uploaded file in the dashboard and recommends published sources where a reference cannot stand (pricing).
What came back from a run
On 27 September 2026 we pasted five real references into Verify references. Four, Pier and colleagues' study among them, came back verified: "The DOI resolves to this record and the title matches." The fifth was Wakefield and colleagues' 1998 Lancet paper, entered with no DOI, which the Lancet retracted in 2010. It came back: "Title, year and first author match a published record. This paper has since been retracted." The totals: four verified, one retracted. Retracted papers in your reference list says what to do next.
We then dropped Kaatz from the Pier entry, as an NSF list must never do. It came back doubtful: "The publisher's record also names Kaatz, who is not in your entry. Check the authors."
A PMCID check for NIH applicants
Verify references also checks each PMCID at PubMed Central against the paper it matched, which is the NIH rule tested for you. The Pier entry in Vancouver form with its real PMCID came back verified. Given PMC1234567, it came back doubtful: "PMC1234567 is another paper's PMCID at PubMed Central (doi: 10.1097/00000658-199503000-00007). Check the PMCID." Given PMC99999999: "PubMed Central holds no paper under PMC99999999. Check the PMCID." To find the right one, use PMC's converter. No tool can say how a panel will score an application.
Quick questions
Do references count toward the page limit for an NIH application?
No. NIH's Bibliography & References Cited attachment has no page limit, though NIH asks for relevant, current references only, and a notice of funding opportunity can set its own rules.
Does NSF limit the References Cited section?
It sets no page limit, but the section may hold citations only, each naming every author in the order published, with title, journal, volume, pages and year.
Do I need a PMCID for every reference in an NIH application?
No. Only articles you wrote or co-wrote that arose from NIH support and fall under the Public Access Policy need a PMCID or NIHMSID; other free online works may carry a PMID or URL.
Do references count in the UKRI word limit?
Usually yes: UKRI's Funding Service counts them in the word count of the question they support. ESRC's research grants call is an exception, with a References question of up to 1,000 words.
Can EdCitation check the references in my grant application?
Yes. EdCitation's free Verify references looks each entry up in the publisher's record, flags retractions and reports an author the record names that your entry leaves out, or a PMCID that belongs to another paper. It writes nothing.
References
- Australian Research Council. (2021a, September 14). Adjustments to the ARC's position on preprints. https://www.arc.gov.au/news-publications/media/communiques/adjustments-arcs-position-preprints
- Australian Research Council. (2021b, October 11). Update: Eligibility in ARC grant applications – Preprints. https://www.arc.gov.au/news-publications/media/feature-articles/update-eligibility-arc-grant-applications-preprints
- Biotechnology and Biological Sciences Research Council. (2026, February 2). BBSRC standard research grant: 2026 round one: Applicant-led mode. UK Research and Innovation. https://www.ukri.org/opportunity/bbsrc-standard-research-grant-2026-round-one-applicant-led-mode/
- Economic and Social Research Council. (2026, May 11). ESRC responsive mode: Research grants round two. UK Research and Innovation. https://www.ukri.org/opportunity/esrc-responsive-mode-research-grants-round-two/
- European Research Council. (2025). Application forms: Starting Grant call (HE ERC StG) (Version 6.0) [PDF]. European Commission. https://ec.europa.eu/info/funding-tenders/opportunities/docs/2021-2027/horizon/temp-form/af/af_he-erc-stg_en.pdf
- European Research Council. (2026). Information for applicants to the Starting and Consolidator Grant 2027 calls (Version 11.0) [PDF]. European Commission. https://ec.europa.eu/info/funding-tenders/opportunities/docs/2021-2027/horizon/guidance/information-for-applicants_he-erc-stg-cog_en.pdf
- National Institutes of Health. (2017, March 24). Reporting preprints and other interim research products (NOT-OD-17-050). https://grants.nih.gov/grants/guide/notice-files/NOT-OD-17-050.html
- National Institutes of Health. (2024a, December 17). 2024 NIH Public Access Policy (NOT-OD-25-047). https://grants.nih.gov/grants/guide/notice-files/NOT-OD-25-047.html
- National Institutes of Health. (2024b). G.220: R&R Other Project Information form. General application guide for NIH and other PHS agencies, FORMS-I. https://grants.nih.gov/grants/how-to-apply-application-guide/forms-i/general/g.220-r&r-other-project-information-form.htm
- National Institutes of Health. (2024c, August 23). Page limits. https://grants.nih.gov/grants-process/write-application/how-to-apply-application-guide/page-limits
- National Institutes of Health. (2025, April 30). Revision: Notice of updated effective date for the 2024 NIH Public Access Policy (NOT-OD-25-101). https://grants.nih.gov/grants/guide/notice-files/NOT-OD-25-101.html
- National Science Foundation. (n.d.). NSF Proposal & Award Policies & Procedures Guide (PAPPG). https://www.nsf.gov/policies/pappg
- National Science Foundation. (2024). Chapter II: Proposal preparation instructions. Proposal and Award Policies and Procedures Guide (NSF 24-1). https://www.nsf.gov/policies/pappg/24-1/ch-2-proposal-preparation
- Pier, E. L., Brauer, M., Filut, A., Kaatz, A., Raclaw, J., Nathan, M. J., Ford, C. E., & Carnes, M. (2018). Low agreement among reviewers evaluating the same NIH grant applications. Proceedings of the National Academy of Sciences, 115(12), 2952–2957. https://doi.org/10.1073/pnas.1714379115
- UK Research and Innovation. (2026, February 19). How applicants use the UKRI Funding Service. https://www.ukri.org/apply-for-funding/develop-your-application/how-applicants-use-the-ukri-funding-service/