Suggested language for citing this research
For the historical series: “Project Selection (2026), independent compilation of Ministry of Defence written answer to Parliament, UK Regular Forces annual average personnel 2000–2023, via MOD Question 10473 answered 25 January 2024.” The original data are MOD public sector information under the Open Government Licence. The analytical visualization and derived comparisons are independent work and not MOD-endorsed.
For the movement comparison: “Project Selection, UK Regular Forces July-ended intake and outflow 2020–2026, transcribed from dated Ministry of Defence Quarterly Service Personnel Statistics releases, retrieved and reviewed October 2026.” Cite the specific GOV.UK release for the reporting year in addition to this research site.
For the modern strength series: “Project Selection, July 2017–2026 comparative UK Armed Forces strength, compiled from three specific historical MOD publication vintages.” Note the 2019 historical revision difference; the time series is an editorial compilation, not one homogeneously reissued MOD table.
Publication checks that protect your article
Before quoting a figure, identify the exact military population and reference date. Read the definitions and note rounding. Separate all-outflow from voluntary exits, and trained voluntary rate percentages from untrained and trained all-Regular exits. Verify whether an MOD statistical revision was published after our review date. A possible change to fitness rules is not evidence that a particular recruitment trend was caused by those changes.
Do not extrapolate acceptance rates by dividing service application records by UK Regular intake. Individuals and application records are not matched cohorts. Do not call Royal Navy/Marines combined figures a Royal Marines-only count. A rounded 2026 net difference should not be “fixed” by altering the official published numbers.
Known limitations we actively disclose
Our 24-year archive stops in calendar 2023 because the official annual-mean answer stops there. We do not append 2024–2026 July snapshots and call them calendar means. The modern July series begins in 2017 and covers three separate underlying statistical source vintages. Our flow archive uses windows ending 30 June, not financial years. The 2000 mean depends in part on interpolated January 2000 data and 2000–2007 observations are less precise than later years.
Supporting questions for an editor
Could the MOD provide one fully revised and methodologically comparable 2000–2026 annual mean dataset? What proportion of 2023–2026 departing Regular Forces left voluntarily, by trained status and service? What did surveys actually say about the reasons for leaving—and how did survey participation and composition change over the same period? These are valuable follow-up reporting questions; the current evidence does not answer them conclusively.
To submit a correction, quote the original source, table, date and column via our contact page. We review changes before republishing and retain an audit trail in version control.