Our objective was to prospectively assess whether early (ie, 2 weeks) response to an antipsychotic predicts later (12-week) response and whether 'switching' early non-responders to another antipsychotic is a better strategy than 'staying'. This randomized, double-blind, flexible-dosed, 12-week study enrolled 628 patients diagnosed with schizophrenia or schizoaffective disorder. All initiated treatment with risperidone. Early response was defined as X20% improvement on the Positive and Negative Syndrome Scale (PANSS) total score following 2 weeks of treatment. Early responders (ERs) continued on risperidone, whereas early non-responders (ENRs) were randomized (1 : 1) to continue on risperidone 2-6 mg/day or switch to olanzapine 10-20 mg/day for 10 additional weeks. Compared with ENRs, risperidone ERs showed significantly greater reduction in PANSS total score (end point; po001). Early response/nonresponse was highly predictive of subsequent clinical outcomes. Switching risperidone ENRs to olanzapine at week 2 resulted in a small but significantly greater reduction in PANSS total score (end point; p ¼ 0.020) and in depressive symptoms (end point; p ¼ 0.004); the reduction in PANSS was greater among those who were still moderately ill at 2 weeks. Switching risperidone ENRs to olanzapine also resulted in significantly greater increases in triglycerides, a significantly greater decrease in prolactin, and significantly less treatmentemergent dyskinesia. This is the first study to prospectively show that early response/non-response to an antipsychotic (risperidone) is a reliable clinical marker of subsequent clinical outcomes and that a 'switching' strategy based on this information may lead to greater clinical improvement than staying on a drug for a longer period in some patients.