A sample of 85 patients with schizophrenia, of whom 34 later dropped out, received randomised treatment. There were no significant differences between treatment-takers and drop-outs in the variables assessed. Patients received either standard-dose maintenance neuroleptic treatment or targeted maintenance pharmacotherapy and all patients received behavioural family therapy. Measures of psychopathology, social adjustment, side-effects, family burden, and expressed emotion were assessed at baseline and then periodically over an 18-month period. The study was designed to compare the two alternative pharmacological maintenance approaches, each of them supported by psychosocial intervention. Any evaluation of the impact of behavioural family treatment on relapse rates and other outcome criteria is exclusively descriptive. A significantly higher rate of relapse was observed at 18 months in patients randomised to targeted treatment compared to those randomised to standard-dose treatment (35% vs 4%). Although patients assigned to the targeted maintenance group received significantly lower mean doses of neuroleptics, there were no significant differences between the two groups with regard to side-effects, global measures of social function, and overall psychopathology. Family burden was higher in the targeted-treatment group at six months, but did not differ at the one-year and eighteen-month time points. However, both groups improved significantly from baseline to 12 or 18 months in almost all variables assessed. Thus, the behavioural family approach did not compensate for the problems associated with the targeted medication strategy.