In a controlled trial, 20 patients at high risk of repeated suicide attempts were randomly allocated to either cognitive-behavioural problem solving or a 'treatment-as-usual' control condition. The group practising problem solving improved significantly more than controls on ratings of depression, hopelessness, suicidal ideation and target problems at the end of treatment and at follow-up of up to one year, and there was evidence of an effect on the rates of repetition over the six months after treatment.