Objective To determine whether the prescription of aripiprazole, compared with olanzapine and haloperidol, was associated with a lower frequency of metabolic syndrome (MS) and treatment discontinuation at 1 year. Method Patients were randomly assigned to be treated open-label and according to usual clinical practice with either aripiprazole, olanzapine, or haloperidol and followed up for 1 year. Results Three hundred out-patients with persistent schizophrenia were recruited in 35 mental health services. The intention-to-treat (ITT) analysis found no significant differences in the rate of MS between aripiprazole (37%), olanzapine (47%), and haloperidol (42%). Treatment discontinuation for any cause was higher for aripiprazole (52%) than for olanzapine (33%; OR, 0.41; P = 0.004), or haloperidol (37%; OR, 0.51; P = 0.030). No significant difference was found between olanzapine and haloperidol. Time to discontinuation for any cause was longer for olanzapine than for aripiprazole (HR, 0.55; P