A preprint server is an online platform where researchers upload scientific papers that have not yet been published in a peer-reviewed journal. These papers, called preprints, are freely accessible to anyone and typically go live within a day of submission. The system exists to get new findings into the public eye quickly, bypassing the months or even years that traditional journal publication can take.
How Preprint Servers Work
When a researcher finishes a study, they have two paths. The traditional route is submitting to a journal, where editors send the paper out for peer review, a process that can stretch from weeks to over a year. The preprint route is much faster: a researcher uploads their manuscript, and it’s usually posted the day of submission or the next day.
That speed comes with a tradeoff. There is no formal peer review before a preprint goes live. However, preprint servers aren’t a free-for-all. About 82% of preprint servers screen submissions before or after making them public, checking for plagiarism, offensive or dangerous content, and whether the work is genuinely scientific in nature. Some servers, like Research Square, use a screening checklist and even award a badge when a paper passes those checks. But none of this replaces the deeper scrutiny of peer review, where experts in the field evaluate a study’s methods, data, and conclusions.
Once posted, each preprint gets a DOI (a permanent digital identifier, the same kind used by journal articles), making it citable from the moment it appears. Authors can upload revised versions as they get feedback or refine their work, and the DOI automatically points to the latest version. If the paper is later published in a journal, the preprint’s record is updated with a link to the final published version, connecting the two permanently.
The Major Preprint Servers by Field
The oldest and most established preprint server is arXiv, launched in 1991 at Los Alamos National Laboratory. Originally created to level the playing field by giving researchers worldwide equal access to the latest physics results, arXiv now covers math, computer science, quantitative biology, and statistics as well. It was the proof of concept that preprints could work at scale.
Biology has bioRxiv, and the health sciences have medRxiv. These two servers, both hosted by Cold Spring Harbor Laboratory, became especially prominent during the COVID-19 pandemic when the urgency of the crisis pushed thousands of researchers to share findings before journals could process them. Other servers cover chemistry, social sciences, engineering, and more specialized niches.
Why Researchers Post Preprints
Speed is the most obvious reason. In fields that move fast, waiting a year for journal publication means your discovery might be scooped. Posting a preprint timestamps your work, establishing priority for a finding on the day you upload it.
Authors also retain copyright of their preprints, unlike many journal articles where copyright transfers to the publisher. This means researchers can share their own work freely without worrying about paywalls. The original purpose of preprints was exactly this: making new scientific knowledge available to the public before the slow process of traditional validation plays out.
Preprint servers also function as a feedback mechanism. Each posted article allows public comments, so researchers can get input from colleagues worldwide before submitting to a journal. That feedback loop can improve the final paper significantly.
The Peer Review Question
The biggest criticism of preprint servers is straightforward: without peer review, inaccurate information can circulate freely. The final version of a paper published in a journal may differ significantly from the preprint, sometimes in ways that change the conclusions entirely. This risk is amplified when journalists or social media users pick up a preprint and present its findings as settled science.
Reputable preprint servers display prominent warnings that the work has not been peer reviewed and should not be used to guide clinical decisions or policy. But those warnings don’t always survive when a headline goes viral. Passing through peer review remains the best assurance that a study’s findings are valid and reliable, even though the process is imperfect in its own ways.
The data on what happens after posting is revealing. A scoping review of health-related preprints found that the median publication rate was 42%, meaning fewer than half of preprints in that field eventually made it into a peer-reviewed journal. The range was wide, from about 22% to 67% across different studies, but the overall picture is that a substantial portion of preprints never go through formal validation.
How Journals Handle Preprints
One practical concern for researchers is whether posting a preprint will disqualify their paper from journal submission. Most major journals have answered this question clearly: preprints are fine. Nature, Science, and Cell all accept manuscripts that have previously appeared on established preprint servers, and many specialty journals have followed suit.
The notable exception is the Ingelfinger Rule, named after a former editor of The New England Journal of Medicine. Under this rule, NEJM considers a preprint to be prior publication, making the manuscript ineligible for submission. This is increasingly a minority position, but it still carries weight in medicine. Science has a more nuanced stance: preprints are acceptable, but if the main findings get heavy coverage in the mass media before publication, the journal may question whether the work is still novel enough to warrant a paper.
For most researchers in most fields, though, posting a preprint and then submitting to a journal is now a standard part of the publishing workflow. The preprint establishes a public record and a timestamp, while the journal article provides the formal validation that hiring committees, funding agencies, and other researchers rely on.
How to Read a Preprint Critically
If you encounter a preprint while searching for health or science information, the key thing to remember is that it sits at an earlier stage than a published journal article. Look for the label: preprint servers clearly mark papers as “not peer reviewed.” Check whether a published version exists by following the DOI or searching for the title in Google Scholar. If a journal version is available, read that one instead.
When no published version exists, treat the findings as preliminary. The methods may have flaws that peer reviewers would catch. The conclusions may be overstated. The results may not replicate. None of this means the preprint is wrong, but it does mean the work hasn’t yet passed through the filter that science relies on to separate solid findings from noise. Preprint servers are a powerful tool for accelerating research, but they work best when everyone involved, researchers and readers alike, understands what “not yet peer reviewed” actually means.