Peer review in numbers
Peer review is a large and mostly unseen part of research. These are published figures about it: how much work it is, who does it, how hard reviewers are to find, how long a decision takes, and what retractions show. Every figure links to its source, and each section says what the figure does not show.
How much work it is
One estimate puts reviewers' time at over 100 million hours in 2020, or more than 15 thousand years, counted across about 21.8 million reviews at six hours each.[1] Valuing that time at researchers' salaries gave over $1.5 billion for reviewers based in the US alone.[1]
What this does not show. It is an estimate, not a measurement. It uses six hours per review and covers only a portion of the world's journals, so the authors describe it as very likely an under-estimate.[1]
Who does the work
In biomedical research, a simulation of the MEDLINE literature found that 20% of researchers performed between 69% and 94% of the reviews.[2] The same study found that the supply of reviewers exceeded demand overall, so the strain comes from how unevenly the work is shared.[2]
The Publons report found a similar pattern in its own records, where about 10% of reviewers were responsible for about 50% of the review records.[3]
What this does not show. The first figure is modelled, and the range reflects the scenarios the authors tested. The Publons figure describes people who chose to record their reviews on that platform, not all reviewers.
Finding reviewers
In Publons' 2016 editor survey, 75% of editors said that finding reviewers and getting them to accept an invitation is the hardest part of their job.[3] Across the platform's data, an editor needed 2.4 invitations for each completed review in 2017, up from 1.9 in 2013.[3]
An editorial in Acta Odontologica Scandinavica describes the same trend over a longer period: two to three reviewers invited per submission in 2006 to 2010, and six to seven in 2019 to 2024.[4] At one medical education journal, the share of invitations accepted peaked at 56.14% in 2020 and was 35.71% in 2024 and 38.58% in 2025, the lowest since the journal began.[5]
When reviewers decline, the most common reasons are that the article is outside their expertise (about 70% of survey respondents) and that they are too busy with their own research (42%). Respondents could give up to two reasons, so the shares do not add to 100.[3]
What this does not show. The journal figures come from one journal each, and the Holmstrup piece is an editorial rather than a measured sample. They show the direction reported by people running journals, and the Publons figures give the wider picture. None of them gives a single number that applies to every journal.
How long it takes
Across 57 health policy journals, the median time to a first decision before peer review was 10 days, to a first decision after peer review 60.5 days, and to a final decision after peer review 198 days. The final decision ranged from 38 to 314 days between journals.[8]
Author-reported data from SciRev gives a view by field. Total review time for accepted papers averaged 12 weeks in medicine and 13 in public health, against 17 weeks across all fields. In medicine, 16% of papers were through review within a month, 71% within three months and 92% within six.[7] In the 2017 analysis of 3,500 author reports, the first response took 8 to 9 weeks in medicine and public health, and 13 weeks on average across fields.[6]
| Field | Average | Within 1 month | Within 3 months | Within 6 months |
|---|---|---|---|---|
| Medicine | 12 weeks | 16% | 71% | 92% |
| Public health | 13 weeks | 14% | 67% | 91% |
The reviewer's own part of that time is shorter. Across Publons data, the median time to complete a review after accepting the invitation was 16.4 days, with a mean of 19.1 days.[3] The rest of a journal's time includes finding reviewers, any further review rounds and the editor's decision.
What this does not show. SciRev timings are reported by authors who chose to submit them, and the data do not cover journals evenly. The health policy figures describe one specialty, not clinical medicine as a whole. The Publons median covers only the stretch after a reviewer has agreed, so these figures measure different stretches of the process and should not be added together.
How far reviewers agree
A meta-analysis of 48 studies covering 19,443 manuscripts, published between 1966 and 2008, found the level of agreement between reviewers of the same manuscript to be low, with a mean intra-class correlation of 0.34 and a mean Cohen's kappa of 0.17.[12]
What this does not show. The studies are older, come from many fields, and measure agreement on different kinds of recommendation. The result says reviewers often reach different judgements. It does not say which judgement is right.
When a paper is retracted
Retraction Watch, which maintains a public database of retractions, described it at the end of 2023 as 45,000 retractions strong and counting, and reported that journals retracted more than 10,000 papers during that year alone.[9]
One publisher group shows how peer review can fail at scale. Wiley and Hindawi announced in April 2023 that about 1,200 more papers would be retracted for compromised peer review, after 511 announced the previous September. The publisher said special issues had been targeted by paper mills, with researcher identities manipulated and content fabricated.[10]
On the causes, a review of 2,047 biomedical and life-science articles that PubMed listed as retracted on 3 May 2012 found that 21.3% were retracted for error and 67.4% for misconduct. Within misconduct, 43.4% were fraud or suspected fraud, 14.2% duplicate publication and 9.8% plagiarism.[11]
What this does not show. These are counts and shares of retractions, not a rate. They do not say what share of all published papers is retracted, and we have not given one because we have not found a figure we could read at its source. The Hindawi case is one publisher and cannot be read as typical. The 2012 review is old, and its authors note that retraction notices often lack the detail needed to find the true cause, so fraud is probably under-counted.[11] The figures also do not separate retractions that followed a peer review failure from those that did not.
Recognition for reviewers
In Publons' 2018 survey, 84.8% of respondents thought institutions should more explicitly require and recognise peer review contributions.[3] Researchers described reviewing as part of the job, as something to reciprocate and as part of the integrity of the literature. In clinical medicine, the most common reason given was to improve their own writing.[3]
What this does not show. The respondents were Publons users who answered a survey, so they may not represent researchers as a whole.
All the figures
| Figure | Value | Source |
|---|---|---|
| Hours reviewers spent on peer review, 2020 | Over 100 million | Aczel et al. 2021 |
| Value of US reviewers' time, 2020 | Over $1.5 billion | Aczel et al. 2021 |
| Share of reviews done by 20% of researchers | 69% to 94% | Kovanis et al. 2016 |
| Share of review records from the most active 10% of reviewers | About 50% | Publons 2018 |
| Editors who say finding reviewers is the hardest part | 75% | Publons 2018 |
| Invitations per completed review, 2013 and 2017 | 1.9 and 2.4 | Publons 2018 |
| Reviewers invited per submission, 2006 to 2010 and 2019 to 2024 | 2 to 3 and 6 to 7 | Holmstrup 2024 |
| Invitations accepted at one journal, 2020 and 2024 | 56.14% and 35.71% | Morley et al. 2025 |
| Median days for a reviewer to complete a review | 16.4 (mean 19.1) | Publons 2018 |
| Median days to a final decision, health policy journals | 198 | Phillips and Horn 2025 |
| Average total review time, medicine and all fields | 12 weeks and 17 weeks | SciRev |
| First response time, medicine and public health | 8 to 9 weeks | Huisman and Smits 2017 |
| Agreement between reviewers, mean ICC and kappa | 0.34 and 0.17 | Bornmann et al. 2010 |
| Papers retracted by journals in 2023 | More than 10,000 | Retraction Watch 2023 |
| Papers retracted for error and for misconduct (2012 sample) | 21.3% and 67.4% | Fang et al. 2012 |
Sources
[1] Aczel B, Szaszi B, Holcombe AO. A billion-dollar donation: estimating the cost of researchers' time spent on peer review. Research Integrity and Peer Review, 2021.
Estimate from public journal data. The authors say it is very likely an under-estimate.
[2] Kovanis M, Porcher R, Ravaud P, Trinquart L. The global burden of journal peer review in the biomedical literature: strong imbalance in the collective enterprise. PLOS ONE, 2016.
Simulation using MEDLINE-indexed biomedical literature, 1990 to 2015.
[3] Publons. Global State of Peer Review. 2018. Survey of more than 11,000 researchers combined with Publons, ScholarOne and Web of Science data.
Industry report. Publons users are a self-selected group of reviewers.
https://publons.com/static/Publons-Global-State-Of-Peer-Review-2018.pdf
[4] Holmstrup P. The peer review process: growing problem in recruiting qualified reviewers. Acta Odontologica Scandinavica, 2024.
A journal editorial. An informed account, not a measured sample.
[5] Morley CP, Prunuske J, Phillips JP, et al. Reviewer engagement trends at a journal: cause for concern. PRiMER, 2025;9:59.
One journal (medical education research), 2,951 invitations, 2017 to July 2025.
[6] Huisman J, Smits J. Duration and quality of the peer review process: the author's perspective. Scientometrics, 2017;113:633-650.
3,500 review experiences reported by authors to SciRev, 2013 to 2016.
[7] SciRev. Analyses: total review time of accepted papers. scirev.org. Read 4 October 2026.
Author-reported timings submitted to a public site. Not a random sample of journals.
[8] Phillips KA, Horn DM. Review and publication times and reporting across journals on health policy. JAMA Network Open, 2025.
Cross-sectional analysis of 57 health policy journals.
[9] Oransky I. The year at Retraction Watch, 2023: Whew! Retraction Watch, 28 December 2023.
Year-end summary from the organisation that maintains the Retraction Watch Database.
https://retractionwatch.com/2023/12/28/the-year-at-retraction-watch-2023-whew/
[10] Kincaid E. Wiley and Hindawi to retract 1,200 more papers for compromised peer review. Retraction Watch, 5 April 2023.
News report on one publisher's announcement. It is a single case, not a measure of the whole literature.
[11] Fang FC, Steen RG, Casadevall A. Misconduct accounts for the majority of retracted scientific publications. Proceedings of the National Academy of Sciences, 2012.
Review of 2,047 biomedical and life-science retractions listed in PubMed on 3 May 2012.
[12] Bornmann L, Mutz R, Daniel H-D. A reliability-generalization study of journal peer reviews: a multilevel meta-analysis of inter-rater reliability and its determinants. PLOS ONE, 2010.
Meta-analysis of 48 studies, 19,443 manuscripts, 1966 to 2008.
Found an error, or know a better source? Write to contact@peerclave.com and we will check it and correct this page.