How many authors per paper? arXiv data by field (2025)
Key numbers
- An arXiv paper submitted in 2025 had 4 authors at the median (mean 6.0), over 284,486 papers.
- 11.1% of 2025 papers had a single author, against 12.5% in 2022.
- Most authors: High-energy physics, experiment (hep-ex), median 7.
- Fewest authors: Mathematical physics (math-ph), median 2.
- Most single-author papers: Economics (econ), 34%.
Counted by SummarizePaper from the metadata of every arXiv paper submitted in 2022 and 2025 (arXiv OAI-PMH, harvested on 21 November 2026).
How many authors does a research paper have?
Across the 284,486 papers submitted to arXiv in 2025, the median paper had 4 authors and the mean was 6.0. The mean is pulled up by large collaborations: 7.6% of papers had 10 authors or more, and 0.27% had 100 or more.
Papers submitted in 2025 by number of authors:
| Authors | Papers | Share |
|---|---|---|
| 1 | 31,765 | 11.2% |
| 2 | 49,555 | 17.4% |
| 3 | 53,857 | 18.9% |
| 4 | 42,737 | 15.0% |
| 5 | 30,584 | 10.8% |
| 6 | 21,980 | 7.7% |
| 7 | 15,051 | 5.3% |
| 8 | 10,355 | 3.6% |
| 9 | 6,873 | 2.4% |
| 10 | 4,774 | 1.7% |
| 11–99 | 16,176 | 5.7% |
| 100 or more | 779 | 0.3% |
Which fields have the most authors per paper?
By the primary category of papers submitted in 2025, ordered by median:
| Field | Papers | Median authors | Mean | Single author | 10 authors or more |
|---|---|---|---|---|---|
| High-energy physics, experiment (hep-ex) | 1,090 | 7 | 181.8 | 13% | 45% |
| Astrophysics (astro-ph) | 18,660 | 5 | 13.0 | 7% | 27% |
| Other physics (physics.*) | 17,437 | 4 | 6.6 | 9% | 12% |
| Condensed matter (cond-mat) | 18,715 | 4 | 5.5 | 7% | 14% |
| Computer science (cs) | 128,835 | 4 | 5.2 | 6% | 7% |
| Quantitative biology (q-bio) | 3,229 | 4 | 5.2 | 10% | 9% |
| Electrical engineering and systems science (eess) | 12,613 | 4 | 4.7 | 4% | 5% |
| Nuclear physics (nucl-ex, nucl-th) | 1,824 | 3 | 13.4 | 14% | 13% |
| General relativity and quantum cosmology (gr-qc) | 4,471 | 3 | 9.2 | 20% | 2% |
| Quantum physics (quant-ph) | 11,890 | 3 | 4.5 | 11% | 6% |
| High-energy physics, phenomenology (hep-ph) | 5,604 | 3 | 3.6 | 15% | 1% |
| Statistics (stat) | 6,946 | 3 | 3.3 | 10% | 1% |
| Nonlinear sciences (nlin) | 945 | 3 | 2.9 | 16% | 1% |
| High-energy physics, theory (hep-th) | 4,694 | 3 | 2.8 | 21% | 1% |
| Quantitative finance (q-fin) | 1,202 | 3 | 2.8 | 24% | 1% |
| Economics (econ) | 2,313 | 2 | 2.4 | 34% | 0% |
| Mathematics (math) | 42,670 | 2 | 2.3 | 29% | 0% |
| Mathematical physics (math-ph) | 1,348 | 2 | 2.3 | 30% | 0% |
Are papers written by more authors than before?
Over all of arXiv, the median went from 3 authors in 2022 to 4 in 2025, the share of papers with 10 authors or more went from 5.6% to 7.6%, and the share of single-author papers from 12.5% to 11.1%. Medians and shares are used here rather than means, which a few papers with thousands of authors can move: one gravitational-wave paper with 1,800 authors weighs as much as 900 two-author papers.
| Field | Median 2022 | Median 2025 | 10+ authors 2022 | 10+ authors 2025 | Single author 2022 | Single author 2025 |
|---|---|---|---|---|---|---|
| Computer science (cs) | 4 | 4 | 3% | 7% | 6% | 6% |
| Mathematics (math) | 2 | 2 | 0% | 0% | 30% | 29% |
| Condensed matter (cond-mat) | 4 | 4 | 12% | 14% | 7% | 7% |
| Astrophysics (astro-ph) | 5 | 5 | 24% | 27% | 7% | 7% |
| Other physics (physics.*) | 4 | 4 | 11% | 12% | 10% | 9% |
| Electrical engineering and systems science (eess) | 4 | 4 | 4% | 5% | 4% | 4% |
| Quantum physics (quant-ph) | 3 | 3 | 5% | 6% | 12% | 11% |
| Statistics (stat) | 3 | 3 | 1% | 1% | 10% | 10% |
| High-energy physics, phenomenology (hep-ph) | 3 | 3 | 3% | 1% | 16% | 15% |
| High-energy physics, theory (hep-th) | 3 | 3 | 2% | 1% | 21% | 21% |
| General relativity and quantum cosmology (gr-qc) | 3 | 3 | 2% | 2% | 21% | 20% |
| Quantitative biology (q-bio) | 4 | 4 | 7% | 9% | 10% | 10% |
| Economics (econ) | 2 | 2 | 0% | 0% | 28% | 34% |
| Nuclear physics (nucl-ex, nucl-th) | 3 | 3 | 17% | 13% | 16% | 14% |
| Mathematical physics (math-ph) | 2 | 2 | 0% | 0% | 33% | 30% |
| Quantitative finance (q-fin) | 3 | 3 | 0% | 1% | 19% | 24% |
| High-energy physics, experiment (hep-ex) | 4 | 7 | 41% | 45% | 18% | 13% |
| Nonlinear sciences (nlin) | 3 | 3 | 0% | 1% | 16% | 16% |
Which papers had the most authors?
The 2025 arXiv papers with the longest author lists:
- arXiv:2507.06261 — 3,435 authors, Computer science (cs)
- arXiv:2501.01495 — 1,819 authors, Astrophysics (astro-ph)
- arXiv:2508.18081 — 1,812 authors, General relativity and quantum cosmology (gr-qc)
- arXiv:2508.18080 — 1,811 authors, General relativity and quantum cosmology (gr-qc)
- arXiv:2508.18083 — 1,808 authors, Astrophysics (astro-ph)
Read any of them in minutes: each link opens a summary on SummarizePaper. To see who works on a topic rather than who signed one paper, Who works on it? maps the teams behind a research topic.
More research publishing data
- arXiv statistics: submissions per year and field
- The fastest-growing research topics
- How many arXiv preprints are published in journals
Methodology and limits
- Papers: every arXiv paper whose identifier starts with 22xx or 25xx (submitted that year), from the arXiv metadata of arXiv's OAI-PMH interface. New versions of a paper do not count again. The yearly totals match arXiv's own monthly submission statistics.
- Authors: the names in the arXiv author list of the latest version. Some large collaborations sign as one name ("ATLAS Collaboration") on arXiv, so high-energy experiment counts are lower than in the journal versions.
- Field: the archive of the primary category; physics.* is "physics (other)", nucl-ex and nucl-th are nuclear physics, hep-lat is counted with hep-th. Fields with fewer than 200 papers are left out of the table, not of the totals.
- The median is the middle paper: half the papers have as many authors or fewer.