Sustained response to antipsychotic therapy is an important outcome measure for patients with psychotic disorders. Placebo control in studies of relapse prevention contributes valuable information yet provokes much debate. This study, using placebo as a control, evaluated olanzapine's efficacy in preventing a psychotic relapse. Participants were stable minimally symptomatic outpatients with schizophrenia or schizoaffective disorder. The study included 4 phases: (1) 4-day to 9-day screening/evaluation (N = 583), (2) 6-week conversion to open-label olanzapine (N = 493; 10-20 mg/d), (3) 8-week stabilization on olanzapine (N = 458; 10-20 mg/d), and (4) 52-week randomized (2:1), double-blind maintenance with olanzapine (N = 224; 10-20 mg/d) or placebo (N = 102). Primary relapse criteria were clinically significant changes in the Brief Psychiatric Rating Scale (BPRS) positive item cluster or rehospitalization due to positive symptoms. Statistical methodology allowed sequential real-time estimation of efficacy across blinded treatment groups and multiple interim analyses, which permitted study termination when efficacy was significantly different between treatments. A significant between-treatment difference emerged 210 days after first patient randomization to double-blind treatment. Thus, 151 (46.3%) of the randomized patients were discontinued early and 34 (10.4%) of the planned patient enrollment were not required. The olanzapine group had a significantly longer time to relapse (P < 0.0001) than the placebo group. The 6-month cumulative estimated relapse rate (Kaplan-Meier) was 5.5% for olanzapine-treated patients versus 55.2% for placebo-treated patients. The design of this study enabled appropriate statistical testing of the primary hypothesis while minimizing exposure of patients to a less effective treatment than olanzapine. In remitted stabilized patients with schizophrenia or schizoaffective disorder, olanzapine demonstrated a positive benefit-to-risk profile in relapse prevention.