A simple method of classifying adolescent self-poisoners into three sub-groups on the basis of the duration of their problems and the presence or absence of behavioural disturbance is described. The three groups differed markedly in terms of factors concerning their families, especially their relationships with their parents, and in their medical and psychiatric histories, the nature of their overdoses and the problems they were facing at the time. The treatment offered to members of the three groups also differed. The method of classification may offer a useful means of distinguishing between subjects in terms of outcome, including subsequent adjustment and repeat attempts.