How we source these figures
Every number in our statistics section comes from one source, retrieved by a script, and never typed in by hand.
The source
The World Health Organization, Global Health Observatory, indicator MH_12 — Age-standardized suicide rates (per 100 000 population). We currently publish figures for 184countries, all from WHO’s 2021 series, retrieved on 4 August 2026.
You can read the raw data yourself: https://ghoapi.azureedge.net/api/MH_12
Why age-standardised
A crude rate is deaths divided by population, which makes a country with an older population look worse for reasons that have nothing to do with risk. Age-standardising adjusts for that so two countries can be compared at all. It is the only form in which a comparison means anything.
Why every figure has a range
WHO publishes an uncertainty interval alongside each estimate, and we render it. A single number implies a precision the underlying data does not have — these are modelled estimates built on national registration systems of very different quality, not headcounts.
Why some countries have no page
Because WHO does not publish a figure for them. We do not estimate, interpolate, borrow a regional average, or carry a number forward from an older year to fill a gap. A country with no defensible figure gets no statistics page — the same rule as our helplines, where a contact nobody has checked is never rendered.
Comparison between countries is also less meaningful than it looks. Where suicide is illegal, or where a family faces religious or insurance consequences, deaths are recorded as accidental or undetermined. A low figure may describe a low rate, or a reluctant coroner.
Why there is no breakdown by method
This is the most important thing on this page, and it is enforced in code rather than promised in prose.
The data structure that holds these statistics has no field capable of containing a method breakdown. Not one that is left empty — one that does not exist. The fetch script accepts only two dimensions, sex and year, and refuses to run at all if the source ever adds a third, rather than quietly discarding it.
That is deliberately stricter than our language rules. A word-level check cannot help here: method data arrives pre-tokenised, as a column heading or a diagnostic code, in a form no blocklist contains. The only reliable defence is to leave no shape for it to occupy.
How this stays current
The figures live in a machine-written file that is committed to the repository, so what you see is exactly what the script retrieved, and the retrieval date is shown on every page. When WHO publishes a new series we re-run the script; nobody edits a number by hand, because a figure somebody typed is a figure nobody can trace.
Found something wrong? Tell us — corrections to data are the most useful thing anyone sends us.