Data on extrapyramidal symptoms (EPS) from both arms of the North American multicenter comparative study of risperidone, placebo, and haloperidol were analyzed. The subjects were 523 patients with chronic schizophrenia who, after a 1-week washout period, received placebo, risperidone (2, 6, 10, or 16 mg/day), or haloperidol (20 mg/day) for 8 weeks; the trial was completed by 253 patients. Severity of EPS was assessed by means of the Extrapyramidal Symptom Rating Scale (ESRS). Mean changes (increases) in ESRS scores from baseline to worst score were significantly lower in each risperidone group than the haloperidol group on the total ESRS (parkinsonism + dystonia + dyskinesia), total parkinsonism, hypokinetic symptoms, and on the questionnaire (p < 0.001). On several of the subscales (dyskinesia, buccolinguomasticatory, and Clinical Global Impression severity of dyskinesia), mean change scores were significantly lower in some of the risperidone groups than in the placebo group (p < 0.05). At the clinically most effective risperidone dose (6 mg/day), the mean ESRS change score was not significantly different from that of the placebo group. A significant linear relationship was noted between mean change scores and increasing risperidone dose on 4 of the 12 ESRS subscales; nevertheless, even at 16 mg/day of risperidone, mean change scores were lower than in the haloperidol group. A linear relationship between increasing risperidone dose and use of antiparkinsonian medications was also apparent. Acute dystonic reactions occurred both in patients receiving risperidone and haloperidol. Patients with severe baseline EPS were at higher risk of EPS during the study than patients with low or moderate baseline EPS. It is concluded that low doses of risperidone cause few or no EPS and recommendations for initiation of risperidone treatment are made.