Preprint Version Numbers: v1/v2/v3 Explained

Most preprint servers let authors post revisions after the initial submission, and each revision gets its own version number — v1, v2, v3, and so on. The catch is that the underlying identifier a citation manager, a reference list, or a journal’s manuscript-tracking system points to does not always change when the version does. Depending on the server, “the preprint” and “this specific version of the preprint” can be the same URL or two different ones, and getting that distinction wrong is how a citation ends up pointing to a different set of claims, figures, or data than the one the citing author actually read. This guide covers how the major preprint servers implement version numbering, how each version gets (or doesn’t get) its own persistent identifier, and how to cite and track versions correctly.

For preprint servers and their general role in scholarly communication, see Preprint Servers Explained. For citation formatting in APA, MLA, Chicago, and Vancouver style, see How to Cite a Preprint — this guide focuses specifically on the version-numbering and identifier mechanics that sit underneath any citation style.

Why preprints carry version numbers

A journal article’s version of record (VoR) is fixed at publication; a preprint is explicitly not. Authors commonly post a revised version after community feedback, after correcting an error, after adding data, or after a journal’s peer review prompts changes that the author also wants reflected on the preprint. Because preprints exist precisely to circulate research before that fixed point, servers need a way to (1) preserve the full revision history rather than silently overwriting earlier claims, and (2) let a reader or citer identify exactly which state of the manuscript they’re looking at. Version numbering is that mechanism. It is a different concept from the broader taxonomy of article versions defined in NISO’s Journal Article Versions (JAV) recommended practice (Author’s Original, Submitted Manuscript Under Review, Accepted Manuscript, Proof, Version of Record) — JAV describes a manuscript’s stages through the *publication* pipeline, while preprint version numbers describe successive revisions *within* the preprint stage, before or independent of any journal submission.

How the major servers implement versioning

arXiv

arXiv assigns version numbers as a literal “v” plus an integer starting at v1. Submitting a replacement increments the version by one; the version identifier takes the form arXiv:YYMM.NNNNNvX. Every version is permanent and stays publicly accessible — the base abstract page (e.g. arxiv.org/abs/2201.NNNNN) always displays the most recent version, but the page’s “Submission history” section lists every prior version with its own stable URL (arxiv.org/abs/2201.NNNNNv1, v2, etc.), and the author-supplied Comments field is often used to summarize what changed between versions. Critically, arXiv does not issue a new DOI for a new version: arXiv’s DOI (10.48550/arXiv.YYMM.NNNNN) resolves to the arXiv identifier’s canonical page, which always shows the latest version, and the DOI’s metadata at DataCite is updated in place rather than replaced with a new record. If you need to point to a specific version rather than “whatever is current,” cite the versioned identifier (arXiv:2201.NNNNNv2) directly, not the bare DOI.

bioRxiv and medRxiv (openRxiv)

bioRxiv and medRxiv, both operated by openRxiv, follow the same versioning model as each other and it differs from arXiv’s in one important respect: openRxiv assigns a single DOI to an article that covers every version, and that DOI always resolves to the latest version — there is no separate DOI per version. Each version still gets its own stable, human-visible URL (for example biorxiv.org/content/10.1101/2019.12.11.123456v2), and prior versions remain publicly accessible through the record’s version history. To cite a specific version rather than “whatever is current,” openRxiv’s own guidance is to append the version to the DOI in the citation text, e.g. https://doi.org/10.1101/2019.12.11.123456 (version 2), or to link directly to the version-specific URL. One further wrinkle worth knowing if you’re checking older citations against current records: openRxiv’s DOI prefix changed from 10.1101 to 10.64898 for newly posted preprints starting December 1, 2025, coinciding with openRxiv’s transition away from Cold Spring Harbor Laboratory’s management — DOIs already assigned under the 10.1101 prefix are unaffected and continue to resolve normally. See How to Submit to bioRxiv or medRxiv for the submission side of this process.

SSRN

SSRN’s versioning model is the loosest of the group. SSRN added DOIs for preprints starting in 2023, primarily to make it easier to link a preprint record forward to its eventual journal version of record. But SSRN’s default revision behavior is to replace the existing paper record when an author submits a revision — the platform’s own support documentation states that previous versions are not automatically saved, and staff cannot restore an earlier revision after the fact. If an author specifically wants multiple versions to remain independently visible (for example, a translated version, a version in a different working-paper series, or a genuinely distinct revision history), they have to opt into SSRN’s “Version Group” feature, which groups the related records under one author page rather than silently overwriting one with another. Practically: unlike arXiv or bioRxiv/medRxiv, you should not assume an older SSRN version you cited is still retrievable at the same URL later — if the exact revision matters, save your own copy (or a web archive snapshot) at the time you cite it.

Preprints.org

Preprints.org (operated by MDPI) also supports posting revised versions of a preprint, numbered sequentially and displayed on the record’s version-history tab, with each version’s PDF retained and individually accessible rather than overwritten. As with the other servers covered here, if you’re citing a specific version, check the record’s version-history tab for the version number and its posting date rather than assuming the version you see by default (typically the latest) is the one referenced in whatever source pointed you to the preprint. See Preprints.org vs. bioRxiv for a broader platform comparison.

Server versioning at a glance

Server	Version-specific URL?	New DOI per version?	Old versions retained by default?
arXiv	Yes — `/abs/IDvN`	No — one DOI resolves to latest; DataCite metadata updated	Yes, permanently
bioRxiv / medRxiv	Yes — `/content/DOIvN`	No — one DOI resolves to latest	Yes, permanently
SSRN	Generally no — revisions replace the record by default	DOI (since 2023) attaches to the paper record, not per revision	No, unless the author opts into a Version Group
Preprints.org	Yes — version-history tab	Not per-version	Yes

This table describes each platform’s own default versioning mechanics as documented by the platforms themselves; always check the specific record’s version history rather than assuming, since individual policies can be updated.

How to cite a specific preprint version

Regardless of citation style, a version-specific preprint citation needs three things a non-versioned citation lacks: the preprint server as the source, an explicit non-peer-review marker, and the version identifier or version-specific URL rather than the server’s general landing page for the work. In practice:

arXiv: cite the full versioned identifier, e.g. arXiv:2201.01234v2, and/or link to arxiv.org/abs/2201.01234v2 rather than the un-versioned abstract page.
bioRxiv/medRxiv: cite the DOI with the version appended (doi.org/10.1101/2019.12.11.123456, version 2) or link to the version-specific content URL.
SSRN: cite the paper’s SSRN ID and, if the exact revision matters, note the “last revised” date shown on the record at the time you accessed it — and keep your own archived copy, since the platform does not guarantee that exact revision remains retrievable later.
Preprints.org: cite the version number as shown on the record’s version-history tab.

For full citation-style formatting (APA, MLA, Chicago, Vancouver), see How to Cite a Preprint. The version-identification practice above applies underneath any of those styles.

How downstream systems track version relationships

Version tracking downstream of the preprint server itself is uneven, which is exactly why pinning the version at the point of citation matters:

Crossref: Crossref metadata supports relation types that link a preprint DOI to its eventual published article (commonly surfaced as “This article is a preprint of” / “This preprint has been published as” on both a preprint’s page and the published article’s page). These relations connect the preprint record to the journal version of record; they are not, on servers like bioRxiv/medRxiv/arXiv, a mechanism for distinguishing v1 from v2 of the same preprint, since those servers use one DOI (or one canonical identifier) across all versions in the first place. Version-to-version tracking is therefore handled by the preprint server’s own metadata (submission history, version-history tab), not by Crossref relation types.
DataCite: for arXiv specifically, DataCite is the DOI registration agency, and arXiv updates the DataCite metadata record in place each time a new version is posted, rather than minting a new DOI.
Journals and editorial systems: when a manuscript that started as a preprint is submitted to a journal, editorial staff typically ask authors to disclose the preprint (per COPE’s guidance on handling material previously posted to a preprint server) and to confirm which preprint version corresponds to the submitted manuscript, since substantive changes between preprint versions can be relevant to peer review.
Reference managers (Zotero, EndNote, Mendeley, etc.): most import preprint metadata via the DOI or identifier at the time of import, which means they capture whichever version was current at import time unless the version is manually recorded in the entry. They do not automatically alert a user if the cited preprint has since been revised.

Practical guidance for authors posting a new version

Use the version-description or comments field (where the server provides one, e.g. arXiv’s Comments field) to summarize what changed — this is often the only human-readable record of what differs between v1 and v2 beyond a diff of the PDFs themselves.
Before posting a substantive revision, consider whether the change affects conclusions readers may have already cited — on servers where old versions remain accessible (arXiv, bioRxiv, medRxiv, Preprints.org), this is less urgent since the earlier version stays retrievable and citable; on SSRN, where a revision replaces the record by default, a significant change effectively erases public access to what was previously cited unless you use SSRN’s Version Group feature.
If the preprint has since been published as a journal article, most servers (and Crossref’s relation-type metadata) support linking the preprint record forward to the published version of record — do this once publication happens so readers who land on the preprint are pointed to the peer-reviewed version.

Practical guidance for citing a specific version

Always resolve the version-specific URL or identifier at the time you cite, not just the server’s general landing page for the work — the landing page typically shows whatever is currently the latest version, which may not be what you actually read.
Re-check the citation shortly before submission. On a fast-moving preprint, a new version can post between when you drafted a citation and when the manuscript is submitted, and reviewers checking the reference may see different content than you cited.
Check whether the preprint has since been published in a peer-reviewed journal. Citing the peer-reviewed version of record, where one now exists, is generally preferable to citing an outdated preprint version, unless you specifically need to reference preprint-stage content that changed or was removed by the time of formal publication.

Frequently asked questions

Does posting a new preprint version change its DOI?

On arXiv, bioRxiv, and medRxiv, no — the DOI stays the same across all versions and resolves to the latest one; the version is tracked through a separate version-specific identifier or URL, not a new DOI. SSRN’s DOI similarly attaches to the paper record as a whole rather than to individual revisions.

If I don’t specify a version, which one am I citing?

By default, a bare DOI or un-versioned URL resolves to whatever is currently the latest version on that server. If the content you’re relying on later changes (a figure is corrected, a conclusion is revised), an un-versioned citation silently points readers to different content than what you actually read.

Can I cite an older version if a newer one now exists?

Generally yes, on servers that retain prior versions (arXiv, bioRxiv, medRxiv, Preprints.org) — cite the specific version identifier or version-specific URL for the version you used. On SSRN, an older version may no longer be retrievable once replaced, unless the author used SSRN’s Version Group feature to keep it independently accessible.

Does a preprint’s version number reset once it’s published in a journal?

No. The preprint’s version numbering is entirely separate from the published article’s identifiers (its own DOI, volume/issue, page numbers). The preprint record and the journal version of record are typically linked to each other via relation metadata, but they remain distinct records with their own version/identifier histories.

Preprint Version Numbers (v1, v2, v3): Citation and Tracking Across Servers