Preprints are research manuscripts shared before conventional journal publication. Most established preprint servers do not post every uploaded file instantly. They screen submissions first. The screen can catch an out-of-scope article, missing ethics information, identifying participant details, duplicate material, or content that could create a serious risk if posted.
That screening is not peer review. A manuscript can pass a server’s checks while still having weak methods, a mistaken analysis, or conclusions that do not follow from the data. Screening answers a limited question: may this manuscript be posted under this server’s rules? Peer review asks a broader scientific and editorial question.
The first screen is usually scope and completeness#
A server begins by deciding whether the submission belongs there. bioRxiv accepts research across the life sciences. medRxiv focuses on health-related research and accepts specified research manuscripts and protocols, not every form of medical writing. A narrative review, opinion article, or case report may therefore be rejected as out of scope even if it is competently written.
Staff also check submission metadata and required fields. Are the authors and group names entered correctly? Is an abstract present? Does the manuscript match the selected article type? Are declarations supplied in the places where the server expects them?
These checks matter because a preprint becomes a public scholarly object. It may receive a persistent identifier, enter search indexes, be cited, and be discussed before any journal decides whether to review it. Basic metadata errors are easier to correct before posting.
Medical preprints receive added risk checks#
medRxiv describes a two-part screening process. In-house screeners first examine scope, article type, submission details, ethics information, possible participant identification, and trial registration where applicable. Automated text analysis also looks for possible plagiarism or material already available elsewhere.
Volunteer clinicians and health professionals then consider whether the manuscript presents health-related research with methods and data and whether posting could cause public harm. A concern can trigger discussion by the content team, founders, affiliates, or outside advisers.
Examples for escalation include an unclear ethics declaration, potentially identifying information, a missing registration for a clinical trial, or a conclusion that could encourage harmful public behavior. Escalation is not a finding of wrongdoing. It is a pause for a closer decision about posting.
This risk screen helps explain why medical preprints may take longer to appear than authors expect. It also explains why an article that fits another repository may not fit medRxiv. Server rules reflect the consequences and norms of the field they serve.
Biology servers use a related but distinct screen#
bioRxiv states that submitted articles receive a basic check for offensive or nonscientific material, plagiarism, possible health risk, and dual-use research concerns. Its scope and risk decisions are not identical to those of medRxiv because the repositories serve overlapping but different research communities.
The server also states plainly that posted articles have not been peer reviewed, edited, or typeset and that posting does not endorse the methods, assumptions, conclusions, or scientific quality. That boundary is central. A screen can decide that a paper is suitable to share without deciding that it is correct.
Readers can compare this boundary with what peer review can and cannot catch. Peer review itself is imperfect, but it usually involves a deeper assessment of design, analysis, reporting, interpretation, and contribution than a preprint posting screen.
What automated checks can and cannot establish#
Text-matching systems can flag passages that resemble prior material or reveal that a manuscript has already appeared online. Such a flag prompts examination. It does not by itself prove plagiarism because standard methods, references, lawful reuse, and earlier versions can produce overlap.
The reverse matters too. No similarity alert does not establish that a study is original, that its data are authentic, or that all sources were credited properly. Image manipulation, fabricated data, inappropriate analysis, and undisclosed conflicts may require different tools, domain knowledge, or institutional investigation.
Screening also cannot reproduce the full journal workflow. It generally does not recruit several independent specialists to evaluate whether the sample, controls, statistical model, sensitivity analyses, and interpretation support the central claim. It does not promise copyediting or final typesetting.
A persistent identifier is not a quality badge#
After approval for posting, a preprint can receive a digital object identifier. The DOI allows readers and machines to identify and cite the work even when a webpage changes. Version information helps distinguish the initial manuscript from later revisions.
Persistence creates responsibility. A citable record can travel quickly through news, policy discussion, grant review, and later scholarship. Authors should therefore label the work accurately and correct it through the server’s versioning or withdrawal procedures when necessary.
The DOI must not be mistaken for certification. It identifies the object. It does not mean that reviewers verified the analysis or that a journal accepted the paper. The same distinction applies when interpreting embargoes and the Ingelfinger rule: dissemination timing and editorial evaluation are related policy questions, not the same judgment.
How authors can prepare for screening#
Read the server’s current scope and submission rules before uploading. Choose the correct article type and provide the required declarations. For research involving people, remove unnecessary identifying details and explain the ethics approval or exemption accurately. Register applicable trials and include the identifier where required.
Check that every author has agreed to posting, that conflicts and funding are disclosed, and that the version does not violate a journal, sponsor, participant-consent, or data-sharing obligation. Review figures, supplementary files, and metadata, not just the main text.
Do not describe screening as peer review in a CV, press release, or public summary. Use “preprint” prominently and preserve the server’s caution label. If a later journal version appears, link the records so readers can find the evaluated article.
How readers can use a screened preprint#
Begin with the label, version, and date. Then ask whether a journal version, formal review, author response, correction, or withdrawal has appeared. Read the methods and results rather than relying on the abstract or social-media summary.
For consequential decisions, seek corroboration. One screened preprint can raise a useful hypothesis or report timely data, but the confidence placed in it should reflect its design, transparency, consistency with other evidence, and review status.
The safest interpretation is exact: screening means the server found the manuscript eligible to post under its current process. It does not mean the server has certified the findings.
References#
- medRxiv, Screening Procedures on medRxiv
- bioRxiv, About bioRxiv
- bioRxiv, What is an unrefereed preprint?
- medRxiv, Understanding DOIs for preprints
- Committee on Publication Ethics, Preprints discussion document
This article is general information about research communication. It does not establish the validity of any preprint or replace scientific, ethics, legal, or editorial review.
Questions and answers
Are preprints checked before they appear online?
Usually yes, but the checks vary by server and are designed mainly for scope, required declarations, privacy, duplication, and risk rather than full scientific evaluation.
Is preprint screening the same as peer review?
No. Screening does not normally ask independent specialists to validate the methods, analysis, interpretation, or importance to the standard expected in journal peer review.
Why can screening take longer for a medical preprint?
A submission may need discussion about ethics oversight, trial registration, participant identification, article type, or possible public harm before posting.
Does a DOI mean that a preprint has passed peer review?
No. A DOI makes the preprint identifiable and citable. It says nothing by itself about editorial certification or scientific validity.
What should a reader check first?
Confirm that the item is a preprint, note its version and date, look for later journal publication or formal reviews, and assess the methods and claims directly.