The definition, exactly
DfE counts an occasional teacher as someone who is not directly employed by the school or local authority, is in school on census day in early November, and is working under a contract or service agreement of fewer than 28 days.
Every clause in that sentence excludes somebody. The November date excludes anyone doing supply work in any other month. The 28-day limit excludes anyone on a term-long maternity cover. The employment clause excludes a school's own directly employed cover staff.
The latest census recorded 20,500 occasional teachers.
Why the number looks large next to vacancies
The same census recorded 1,600 advertised teacher vacancies, or 4 per 1,000 teachers in service, down from 2,200 the year before.
So on one day in November, schools had substantially more short-contract external staff present than they had posts advertised. That is not a contradiction. Cover and vacancy are different things: most occasional teachers on any given day are covering absence, training or other short gaps, not filling advertised posts.
What it is good for
Tracked consistently over time on the same definition and the same date, it is a genuine indicator of how much short-contract external staffing schools are using. That is worth knowing, and it is measured the same way each year.
What it cannot do is tell you how many people work as supply teachers in England. Supply work is intermittent by nature, so a single-day snapshot will always capture a fraction of the people who do it across a year. There is no published headcount of the supply workforce, and this number is not a substitute for one.
Why this matters
The figure is regularly quoted as though it were the size of the supply teaching workforce. It is not, and the gap between the two is large and unquantified.
For school leaders and anyone reading workforce coverage, the practical rule is to treat it as a measure of census-day temporary presence. Used that way it is informative. Used as a workforce headcount it is simply wrong.