Background. Reports of adolescent suicidal behaviour have generally derived from clinical settings but population-based studies are likely to provide a clearer epidemiological view.

Methods. Non-fatal suicidal behaviours were studied in 1699 Australian 15- to 16-year-old secondary school students at 44 schools in the state of Victoria, Australia. Self-reported episodes of self-harm were characterized using items from the Beck Suicide Intent Scale.

Results. The 12 month weighted prevalence estimate for deliberate self-harm was 5·1%. The commonest forms were self-laceration (1·7%), self-poisoning (1·5%) and deliberate recklessness (1·8%). Self-poisoning and self-laceration were commoner in girls. The prevalence of ‘true suicide attempts’ was 0·2%. Most self-harmers did not perceive death as likely, plan self-harming episodes at length or inform others of the episodes. Psychiatric morbidity had the strongest association with self-harm, an association which held for all subtypes. Antisocial behaviour and substance abuse were associated with self-harm in girls but not boys. Sexual activity was independently associated with self-harm in both genders.

Conclusions. Deliberate self-harm was common but the great majority of episodes were not ‘true suicide attempts’. It is, therefore, possible that attributable mortality and morbidity may be greater in self-harmers without definite suicidal intent.