A preprint server is an online platform where researchers post a manuscript publicly — usually a link to a PDF, a timestamp, and a DOI — before, or entirely separate from, formal peer review at a journal. The core mechanic is the same across every server: the author uploads a manuscript, the server runs a brief, non-peer-review screening pass, and the paper becomes publicly readable, typically within a day or two. This guide covers what that mechanism actually involves: why researchers use preprint servers, how the screening/moderation step differs from peer review, how the major discipline-specific servers compare, how DOIs and versioning work, how preprints get indexed and discovered, and how a preprint eventually links to its published version. For the underlying concept itself, see CASRAI’s definition of preprint; for choosing among servers once you understand the basics, see How to Choose a Preprint Server.
What a Preprint Server Actually Is
A preprint server is infrastructure, not a publisher. It hosts a manuscript version — the author’s own text, figures, and data as submitted — and makes it publicly accessible with a persistent identifier, but it does not evaluate whether the science is correct, significant, or ready for a journal decision. That distinction matters because it’s the single most common point of confusion about preprints: posting to a server is not the same event as publishing in a peer-reviewed journal, even though both processes end with a publicly citable document. A comparison of the two processes side by side is covered in Preprints vs. Peer-Reviewed Articles.
Preprint servers are typically run by nonprofits, university libraries, scholarly societies, or (in a few cases) publishers, and nearly all of the major disciplinary servers are free for authors to deposit to — funded instead through a mix of institutional membership fees, foundation grants, and, in some cases, independent nonprofit governance structures.
Why Researchers Use Preprint Servers
The reasons authors post preprints are consistent across fields, even though preprint adoption rates vary sharply by discipline:
- Speed. Formal peer review and production can take months; a preprint is publicly readable within days of submission, which matters most for fast-moving or time-sensitive fields.
- A timestamped priority claim. A preprint server assigns a dated, citable DOI at the moment of posting, which functions as public, verifiable evidence of when a result was first reported — the same logic that led physics, mathematics, and computer science to normalize preprinting decades before most other fields.
- Open access, unconditionally. A preprint is free to read on posting, regardless of where (or whether) the work is later formally published, or what access model that journal uses.
- Community feedback before formal review. Preprints are visible to the whole field, not just two or three assigned reviewers, so authors sometimes get substantive feedback — or catch an error — before formal submission.
- Funder and mandate compliance. Some funders and institutional policies now explicitly encourage or require preprinting as part of open-science compliance; the NIH, for example, allows preprints to be cited in grant applications and progress reports as evidence of productivity.
How Submission, Screening, and Moderation Actually Work
This is the step most often misunderstood, so it’s worth being precise: preprint server screening is not peer review, and it is not designed to be. Peer review evaluates whether the methodology is sound, the conclusions are supported by the data, and the work meets a journal’s standard for significance and rigor — a process that typically takes weeks to months and involves subject-matter experts assigned specifically to that manuscript. Preprint server screening is a much lighter, faster check that exists to filter out non-scholarly material, plagiarism, and content that poses a public-harm risk — not to validate the science.
The exact process varies by server, but two well-documented examples illustrate the general shape:
- bioRxiv and medRxiv run a two-stage screen: in-house staff first check for completeness, correctly coded metadata, and run an automated plagiarism/text-overlap check (via Similarity Check), then a distributed group of subject-expert “Affiliates” — practicing scientists with principal-investigator-level standing — review each submission to confirm it presents genuine biological or health research and flag any potential for public harm (for example, content with biosecurity or clinical-safety implications) before it posts. Neither stage evaluates whether the methodology or conclusions are correct.
- arXiv uses a moderation and endorsement system: submissions are checked for topical scope and formatting by category moderators, and first-time submitters in most categories need an “endorsement” from an existing arXiv author, a mechanism designed to filter obviously non-scholarly submissions rather than assess quality. See arXiv Preprints: What They Are and How to Use Them for the full mechanics of arXiv’s category and moderation system, and What Is bioRxiv? Submission Guide for a step-by-step walkthrough of submitting there.
Because screening is deliberately light, credible preprint servers compensate with clear labelling: ASAPbio‘s and the Committee on Publication Ethics (COPE)‘s guidance both call for every preprint to be explicitly and visibly marked “not peer reviewed” on the page itself, not buried in a footer. The International Committee of Medical Journal Editors (ICMJE) sets out a similar baseline expectation for what a credible preprint server should do — clear non-peer-review labelling, a documented screening process, a way for readers to flag concerns, and a documented withdrawal process. The full checklist is walked through in How to Choose a Preprint Server. For the specific consequence of this distinction — what happens when a preprinted manuscript is later submitted to a journal — see Preprint Server Policies: Avoiding a “Prior Publication” Conflict and, from the editor’s side, COPE’s Position for Editors on material previously posted to a preprint server.
Major Preprint Servers, by Discipline
There is no single universal preprint server — coverage is split by field, and the servers differ in scope, screening emphasis, and governance:
- arXiv — physics, mathematics, computer science, quantitative biology, quantitative finance, statistics, and electrical engineering/systems science. The oldest and largest preprint server, in continuous operation since 1991.
- bioRxiv — biology and life sciences. Operated by Cold Spring Harbor Laboratory.
- medRxiv — clinical and health sciences, including studies involving human participants, which brings additional screening emphasis on patient-safety and ethical-disclosure concerns given the direct clinical relevance of the content. Co-operated by Cold Spring Harbor Laboratory, Yale, and BMJ.
- SSRN (Social Science Research Network) — economics, finance, law, and other social sciences; also hosts working papers, not only preprints in the strict sense.
- ChemRxiv — chemistry, operated by a partnership of chemical societies.
- PsyArXiv and SocArXiv — psychology and sociology, both built on the OSF Preprints network (see OSF Preprints), a shared open-source infrastructure for discipline-specific community servers.
- EarthArXiv — earth and planetary science. Originally launched on the OSF Preprints network, it moved to an independent, California Digital Library-hosted Janeway platform in October 2020 and is no longer part of the OSF federation. See What Is EarthArXiv? for the full history.
- Research Square and Preprints.org — multidisciplinary servers that accept submissions across fields rather than specializing in one; Research Square also operates “In Review,” a preprint-posting option integrated directly into some journals’ submission systems.
CASRAI maintains direct head-to-head comparisons for the pairings researchers ask about most: arXiv vs. bioRxiv, bioRxiv vs. medRxiv, and arXiv vs. SSRN. For a maintained, comprehensive list of servers by discipline and policy, ASAPbio’s preprint server list is the standing community reference — more current than any static list reproduced here.
DOIs and Versioning: How a Preprint Is Identified Over Time
Every major preprint server assigns a persistent digital object identifier (DOI) at the moment of posting, generally through DataCite or Crossref, which is what makes the “timestamped priority claim” benefit above actually enforceable — the DOI resolves permanently to that specific version, with a registered creation date. Most servers also support explicit versioning: when an author revises a posted manuscript, the server does not overwrite the original but publishes a new numbered version (v1, v2, v3, and so on) while keeping every prior version publicly accessible and separately citable. This preserves the full history of a preprint rather than silently replacing it, which matters both for reproducibility and for resolving disputes about what a given citation actually pointed to at the time.
How Preprints Get Indexed and Discovered
Preprints are broadly discoverable through general academic search rather than through the specialized indexes that catalog peer-reviewed journal literature. Google Scholar indexes preprints from most major servers directly, and Crossref’s metadata (since preprint DOIs are commonly registered through it or through DataCite) feeds preprints into any discovery tool built on Crossref data. Discipline-specific aggregators add further reach — for example, arXiv content flows into NASA ADS and Semantic Scholar, and bioRxiv/medRxiv content is indexed by Europe PMC. Indexing in PubMed itself is narrower and conditional: bioRxiv and medRxiv preprints are only pulled into PubMed when they report NIH-funded research, as part of a specific pilot arrangement, not as a blanket policy covering every preprint on those servers.
From Preprint to Published Version: How the Two Get Linked
When a preprinted manuscript is later formally published — often after substantive changes from peer review — credible preprint servers update the preprint’s own page with a link forward to the published version, and vice versa where the journal supports it. ICMJE’s recommendations make this an explicit expectation of a credible preprint server: authors should be able to link the preprint to its eventual published article, and the preprint record itself should be updated rather than left to stand alone once a formal version of record exists. This linkage is also central to why editorial bodies like COPE treat a disclosed preprint differently from other forms of prior publication — see COPE’s guidance for editors for the full reasoning. It’s also the mechanism that makes citation practice for preprints workable in the first place; see How to Cite a Preprint for format-specific guidance once you need to reference one directly, and Reviewed Preprints: How eLife’s Peer Review Model Works for a case where the preprint-to-published-version relationship has been restructured entirely, rather than simply linked after the fact.
Posting a preprint is not without real trade-offs, and understanding the mechanics above is only half the decision — see 5 Disadvantages of Preprints Every Author Should Weigh for the other side of that assessment, and CASRAI’s scholarly publishing hub for the broader landscape this fits into.
Frequently Asked Questions
Are preprints peer reviewed?
No. A preprint has gone through a server’s screening process — checking for scope, completeness, plagiarism, and public-harm risk — but not through peer review, which evaluates methodological soundness and whether the data support the conclusions. Credible servers label every preprint as “not peer reviewed” specifically to prevent this being assumed.
Do preprints have DOIs?
Yes. Every major preprint server assigns a persistent DOI at posting, typically via DataCite or Crossref, which is what makes preprints citable and gives them a verifiable, timestamped record.
Is posting a preprint considered “prior publication” that would disqualify a journal submission?
No, not for the great majority of journals and fields today. See Preprint Server Policies: Avoiding a “Prior Publication” Conflict for the full explanation and the narrow exceptions that still exist.
Are preprint servers free to use?
Essentially every major disciplinary preprint server is free for authors to deposit to. They’re funded through institutional membership, foundation grants, or nonprofit governance structures rather than author-facing charges.
Can a preprint be cited in a paper or grant application?
Yes, and it’s increasingly common — many funders, including the NIH, explicitly allow preprints to be cited as evidence of productivity in applications and progress reports. See How to Cite a Preprint for citation-format specifics.
