Why application divided by intake is not an acceptance rate
A valid acceptance-rate calculation would require a precisely defined cohort of distinct applicants, a fully known decision outcome for each person, and compatible recruitment stages and reference periods. UK military application statistics are often counts of submitted applications, can represent repeat applications and vary across the services' technical systems and definitions. MOD reports explicitly warn against combining the three services' application figures into one Armed Forces total.
Even if a report states 100,000 applications to a service in one year and 10,000 people entering that year, dividing 10,000 by 100,000 would not establish a 10% chance of selection. A person can apply in one period and enter in another; people can withdraw or change role, submit multiple applications or still be awaiting a decision. Entrant counts can include different recruitment pathways and rejoiners. Neither quantity necessarily refers to the same group.
Strength is a stock; intake and outflow are flows
Imagine a fictional organisation with 500 employees on 1 July, 75 entrants and 70 leavers over the following year. Its size could change by five if no other status changes occur. But other movements and measurement boundaries may affect the recorded size. The arithmetic difference in two reported stock totals is not in general the same statistic as the published intake-minus-outflow figure. In the military, full-time, Regular, Gurkha, Reserve and trained categories require particular care.
Our July 2017–2026 strength charts use point-in-time data. The seven-year flow and voluntary-outflow archive instead records 12-month movement windows ending in June. The 2000–2023 historical series is a third construction: an official calendar-year mean of four snapshots.
What does a voluntary outflow rate actually divide?
The MOD labels a specific subset as Voluntary Outflow (VO): trained Royal Navy/Marines and RAF personnel and trade-trained Army personnel leaving voluntarily before their agreed service end. The published VO rate uses the average appropriate trained/trade-trained personnel population across the reference window. It does not divide all military leavers by all personnel currently serving and it is not a statistical estimate of an individual recruit's chance of resigning.
A 5% trained voluntary-outflow rate must not be presented as “one in twenty applicants quits”. The numerator is those who actually served and exited under the specified definition, not people considering joining.
Why the MoD text sometimes differs from the subtraction
Sources may print 12-month joining and leaving totals rounded to the nearest ten, then describe a net movement calculated from underlying values. For June 2026, 14,500 minus 13,050 equals 1,450, while the MOD says net +1,460. Our series publishes both statements and labels the difference. Altering the official printed numbers to force a match would be less rigorous than acknowledging precision and revisions.
Why Royal Marines cannot always be separated from the Navy
Many high-level official MOD statistical tables combine Royal Navy and Royal Marines under one branch. A combined figure does not reveal the separate strength, entry rate or retention performance of a Commando unit. If a journalist needs Royal Marines-only statistics, they must find an appropriate explicitly defined table or request further data rather than silently assigning a share.
What retention research still needs
There is a difference between counting departures and knowing why personnel leave. The MOD's Continuous Attitude Survey describes experiences and expressed intentions, but response patterns, sampling and survey wording must be examined carefully. Trends in recruitment/retention, pay, housing or workload do not by themselves prove causation. Our research pages are descriptive and openly state these limits.
A practical checklist before quoting any military statistic
State the service and population. Specify whether the period is a date, calendar average, rolling twelve months or financial year. Record the source publication and revision date, and whether counts were rounded. Distinguish an observed value from a derived sum or percentage. Check changes in collection systems and trained definitions before making historical comparisons. Finally, do not describe any aggregate national metric as an individual's selection probability.
All analysis is independent and carries no Ministry of Defence endorsement. Corrections can be submitted via our contact page, with a source URL and table reference.