How Journal Clock measures journal speed

What is measured

For each article, publishers deposit with PubMed the date the manuscript was received, often a revised date, and the date it was accepted. PubMed also records when the article was published online. From these, Journal Clock calculates for each article:

• Submission → acceptance = accepted date − received date, in days.
• Acceptance → online publication = electronic publication date − accepted date.
• Submission → online publication = electronic publication date − received date.

Articles are grouped by the year they were accepted. For each journal and year we report the median, the 25th and 75th percentiles (the middle half of articles), and the 90th percentile when there are at least 200 articles. Percentiles use linear interpolation between ordered values (as Excel's PERCENTILE.INC and R's default) and are rounded to the nearest day.

These figures describe accepted articles only. Rejected manuscripts never appear in public data, so nothing here tells you acceptance rates or how long a rejection takes.

Which articles count

• Every PubMed record is collected, one day of PubMed entry at a time from January 1, 2020, and each day is checked to contain every record PubMed lists for it.
• Records of these publication types are excluded, because they are not peer-reviewed research or were never accepted by the journal: Address, Autobiography, Bibliography, Biography, Clinical Conference, Comment, Conference Proceedings, Congress, Consensus Development Conference, Consensus Development Conference, NIH, Corrected and Republished Article, Dataset, Directory, Duplicate Publication, Editorial, Expression of Concern, Festschrift, Historical Article, Interview, Introductory Journal Article, Lecture, Legal Case, Legislation, Letter, News, Newspaper Article, Obituary, Patient Education Handout, Periodical Index, Personal Narrative, Portrait, Preprint, Published Erratum, Retraction Notice, Retraction of Publication, Video-Audio Media, Webcast.
• Research articles, reviews (Meta-Analysis, Review, Scoping Review, Systematic Review), case reports and other journal articles are included.
• An interval is dropped when a date is missing, when acceptance comes before submission, when submission to acceptance exceeds 1,460 days (four years), when acceptance to online publication exceeds 730 days, or when a date lies in the future. A non-research item accepted on the day it was received is treated as invited content and dropped.
• Journal pages list how many records were excluded for each reason.

Coverage

For every journal and year we ask PubMed separately how many records it holds, and compare that with the records we collected.

• Figures that depend on complete article counts — retractions and retraction rates, publication volume and its change, the share of very fast acceptances, and distributions — are shown only when we collected at least 90% of the journal's records for every year involved. Otherwise the page says “Not computed”, never zero.
• Speed figures for a year require at least 98% of the records for that year and for the following year, because articles accepted late in a year enter PubMed the next year.

3,409 journals meet every rule for 2025. By primary field:

Checks for unreliable dates

Some publishers deposit placeholder or rounded dates. For each journal and year with at least 50 articles, a year is withheld when:

• more than 20% of submission or acceptance dates fall on the 1st of the month (about 3% would by chance);
• more than 25% of research articles show acceptance on the day of submission;
• more than 25% of articles share one acceptance date.

Validation against publisher pages

We checked 200 randomly selected articles against what the publisher itself says the dates were, on Sep 16, 2026: 100.0% matched (153 of 153 articles where the publisher's dates could be read; a match means both the submission and the acceptance date are identical). The dates came from the article page itself for 118 of them and from the publisher's own Crossref metadata for 35, where the page shows no dates. For the remaining 47 the publisher shows no dates we could read, so they are not counted either way. Download the sample, including which source each row was checked against.

When a year is reported

A year is reported only when it ended at least 180 days before the data date and at least as long ago as the time within which 90% of that journal's previous-year articles reached PubMed after acceptance. A median is shown only with at least 50 articles. Tables and sentences headed with a year contain only that year's figures; journals whose latest usable year is earlier are shown separately.

Faster or slower?

The latest reported year is compared with the year two before it. The difference in medians gets a 90% confidence interval from distribution-free standard errors of each median (Price & Bonett, 2002). A journal is labelled faster or slower only when the interval excludes zero and the change is at least 10 days; otherwise “no clear change”.

Fees

• Listed APC: from DOAJ for journals listed there; otherwise the amount OpenAlex reports. Other currencies are converted to US dollars at the latest ECB reference rates (Sep 15, 2026).
• Paid APC: the median of payments institutions reported to OpenAPC over the last three reported years, converted from euros at each year's average ECB rate. Shown only with at least 5 payments.

Integrity and indexing signals

• Retractions: articles in our PubMed data for the journal over 5 years whose PubMed ID or DOI appears as a retraction in the Retraction Watch Database. The rate per 1,000 is shown only with at least 200 articles. Recent articles have had less time to be retracted.
• Accepted within 14 days: thorough peer review rarely finishes in two weeks, so a large share can merit questions — but letters, rapid-communication formats and resubmissions explain many cases.
• MEDLINE: whether any of the journal's articles in its latest year were indexed for MEDLINE.
• Journal Clock never accuses a journal of misconduct and never ranks journals by quality.

Neutrality

No journal or publisher can pay to change, remove or improve its figures. Organisations may buy reports about their own journal; the public numbers stay the same for everyone.

Matching journals

PubMed journal records are matched to OpenAlex by ISSN and grouped under the linking ISSN (ISSN-L). A PubMed journal whose ISSNs point to two different journals is left out rather than guessed (22 at the last build); 0 measurable PubMed journals had no OpenAlex match. Fields come from the topics of each journal's articles in OpenAlex: a journal is listed in a field when at least 25% of its topics fall there.

Corrections

Email rafiq@gradsummit.com or use the contact form with the journal name and ISSN-L. Verified metadata errors are corrected within five business days and logged below.

No corrections have been needed so far.

Sources and terms

Limitations

• Past times are not a prediction for your manuscript.
• Publishers differ in what they record as “received”; some use the date of the revised submission.
• Journals outside PubMed are not covered.