Choosing a preprint server is a different question from knowing what a preprint is. This guide assumes you already know the basics (see the CASRAI Dictionary’s definition of preprint if not) and focuses on the operational decision: which server to post to, what to check before you do, and — the part most “which preprint server” guides skip — how that choice affects the identifiers and metadata your institution’s research information system will need later, when the preprint becomes a published output.
What to check before you post anywhere
Discipline fit matters, but it is a filter, not the whole decision. The International Committee of Medical Journal Editors (ICMJE) sets out what a credible preprint server should do, and it is a reasonable checklist regardless of field:
- Clear non-peer-review labelling. The server must clearly identify posted content as not peer reviewed, on the record itself, not just in a footer disclaimer.
- A documented screening process. Reputable servers screen for non-scholarly material, plagiarism/text overlap, and — where relevant — content with clinical, public health, or biosecurity risk before posting. The Committee on Publication Ethics (COPE) recommends this minimum-screening standard explicitly, precisely because preprint servers vary widely in how much they check.
- A concern-reporting mechanism. A visible way for readers to flag problems with a posted preprint, per ICMJE.
- Withdrawal and version handling. A documented process for withdrawing a preprint, with the reason and date preserved in the metadata rather than the record simply disappearing.
- A mechanism to link the preprint to its eventual published version. This is the criterion most selection guides skip, and it is the one that matters most for research administration — see below.
Disciplinary scope, cost, and interface polish are the criteria most “which preprint server” guides lead with. They are real, but secondary to the above — and the honest answer on cost is that essentially every major disciplinary preprint server (arXiv, bioRxiv, medRxiv, ChemRxiv, SSRN, PsyArXiv, EarthArXiv, Research Square, and the OSF Preprints network) is free for authors to deposit to. They are funded by a mix of academic-institution membership fees, foundation grants, and (in arXiv’s case as of July 2026) an independent nonprofit funding model — not author charges. If you are specifically weighing free preprint server options, cost is rarely the differentiator; screening standards, discipline fit, and journal compatibility are. For a maintained, comparative list of servers by discipline and policy — more current than anything worth reproducing here — ASAPbio’s preprint server list is the standing reference the scholarly-communication community itself uses.
Licensing and funder requirements
Most preprint servers default to, or require, a Creative Commons license (commonly CC-BY) rather than an all-rights-reserved posting. Check the specific server’s license options before posting if your funder has requirements: cOAlition S (the funders behind Plan S) strongly encourages early sharing of results via preprints as part of its 2026-2030 strategy, though — as of this update — that is framed as encouragement rather than a strict deposit mandate. Confirm the current position with your specific funder rather than assuming a blanket requirement either way.
DOIs, versioning, and the link to the published record
This is the part of preprint-server selection that a generic “which server should I use” answer will not cover, and it is the part that actually determines whether your preprint integrates cleanly with research-administration infrastructure later.
Reputable preprint servers register each preprint with Crossref as posted content, a distinct record type from a journal article — not registered as a data object or article. Two relationship types in that metadata matter operationally:
isVersionOflinks successive preprint versions (v1, v2, v3) together, so a reader or system landing on any version can trace the full revision history.isPreprintOflinks the preprint DOI to the DOI of the eventual peer-reviewed version once it publishes. The preprint server (or, in practice, the publisher) has to actually register this relationship — it does not happen automatically just because the same paper exists in both places.
Why this matters beyond tidiness: institutional repositories, CRIS platforms, and funder reporting systems that harvest metadata by DOI depend on this relationship existing to avoid counting the preprint and the published article as two unrelated outputs, or missing the link between them entirely. If you are choosing between two otherwise comparable servers, one that reliably registers isPreprintOf relationships once the paper is published is doing real, if invisible, work for you later.
ORCID, and how the preprint reaches your record automatically
If a preprint server includes an author ORCID iD in the metadata it sends to Crossref at DOI registration, Crossref can notify that author (via their ORCID iD inbox) and, with standing permission granted, add the work to their ORCID record automatically through Crossref’s auto-update service — the same mechanism that updates an ORCID record when a journal article is published. Whether a preprint server supports ORCID collection at submission is a genuinely useful selection criterion if you want your preprint to show up on your ORCID record without manual entry, and it is rarely mentioned in general preprint-server-choice advice.
Where this fits in your institution’s CRIS
For research offices rather than individual authors: a preprint with a Crossref DOI and correct ORCID metadata is already machine-discoverable by the same harvesting routes your CRIS (current research information system — Pure, Symplectic Elements, or similar) uses for published articles. It will not be indexed everywhere a journal article is (Google Scholar and Europe PMC index most major preprint servers; broader indexes vary), but the identifier chain — ORCID, DOI, and, where a funded project is being tracked end-to-end, a RAiD (Research Activity Identifier, standardized as ISO 23527:2022) — is what lets a preprint be correctly attributed to a researcher, a project, and eventually its published version without manual reconciliation. This is the interoperability layer a best-preprint-server ranking never touches, and it is the reason server selection is a research-administration question, not just an author-convenience one.
Check journal compatibility before you post
Most journals now permit prior preprint posting, but policies vary, and ICMJE’s own recommendation is that authors inform the journal of a preprint and provide a link to it at submission, rather than assume permission. Open Policy Finder (the Jisc service that absorbed Sherpa Romeo, Sherpa Fact, and Sherpa Juliet into one platform in November 2024) is the standard place to check a specific journal’s self-archiving and preprint policy before you post, rather than after.
A note on medRxiv and clinical/health preprints
Health-related preprints carry a distinct risk profile — a flawed clinical or public-health finding can influence patient care or policy before peer review catches an error. medRxiv applies additional screening for content with direct clinical implications and requires authors to include a plain-language statement that findings are preliminary and should not guide clinical practice. Since March 2025, medRxiv and its sister server bioRxiv are governed by openRxiv, an independent nonprofit that took over from Cold Spring Harbor Laboratory (medRxiv remains a partnership between openRxiv, BMJ, and Yale University) — a genuine governance shift worth knowing if you are assessing a server’s institutional stability, not just its submission form.
Platform stability generally is worth checking, not assuming: arXiv itself became an independent nonprofit on 1 July 2026, after 25 years hosted by Cornell University, governed now by a board including Cornell and the Simons Foundation as founding members. A server changing its institutional host does not necessarily change anything for authors day to day, but it is exactly the kind of change worth knowing about before you build a submission workflow around a specific platform’s assumed permanence.
Frequently asked questions
What is the best preprint server?
There is not a single best server — the right one is the one that matches your discipline, meets the screening/identification/versioning standards above, and is compatible with your target journal’s policy. arXiv (physics, mathematics, computer science), bioRxiv (life sciences), medRxiv (clinical and health research), ChemRxiv (chemistry), SSRN (social sciences, economics, law), PsyArXiv and EarthArXiv (via the OSF Preprints network), and Research Square (multidisciplinary) each serve different communities well within their scope. ASAPbio’s list is the best-maintained comparison if you need to check policy details for a server not covered here.
Is posting to a preprint server free?
For essentially every major disciplinary preprint server, yes — deposit is free to authors. These platforms are funded by academic institutions, foundations, and (increasingly, following arXiv and bioRxiv/medRxiv’s moves to independent nonprofit governance) diversified nonprofit funding models, not by author-facing charges.
Where can I find a list of preprint servers?
ASAPbio maintains a continuously updated, comparative list of preprint servers with their disciplinary scope, screening practices, and licensing policies — a more reliable reference than a static list on any single guide, since server policies and even ownership (see medRxiv/bioRxiv and arXiv above) do change.
What is medRxiv?
medRxiv is the preprint server for clinical and health-related research, launched in 2019 and now governed by openRxiv in partnership with BMJ and Yale University. It applies additional screening to content with direct clinical or public-health implications and requires a preliminary-findings disclaimer on every posted preprint, given the direct risk of premature clinical or policy action on unreviewed findings.
Does posting a preprint count as prior publication and block journal submission?
Under ICMJE recommendations, no — a preprint is not considered prior publication that would disqualify a manuscript from journal submission. Authors are expected to disclose the preprint and link to it at submission, and, once the paper is accepted, to have the preprint metadata amended to point to the published version.
