OBJECTIVE To compare the risk of extrapyramidal syndromes (EPS) between patients using risperidone and those using low-potency conventional antipsychotic drugs (APDs) in outpatient clinical practice, as measured by the use of anticholinergic medication. We tried to replicate results from previous clinical trials that compared risperidone with high-potency APDs. METHODS Data was obtained from the PHARMO database containing filled prescriptions of 450,000 community-dwelling people in The Netherlands from 1986 to 1998. From the patients aged 15-54 years who had been newly treated with APDs, we defined mutually exclusive cohorts according to the APD first prescribed to a patient. APD exposure was followed until the first prescription of anticholinergic medication and was censored when APD prescribing was interrupted or switched. We estimated relative risks between risperidone and commonly used low-potency and high-potency APDs using Cox proportional hazards models, adjusting for age, gender, dose and other potential confounders. RESULTS In 4094 patients who had been newly prescribed antipsychotic drugs, the overall incidence rate of anticholinergic drug therapy was 556 per 1000 person-years, which was dose dependent. Prescribed doses of all antipsychotics were low. While, in accordance with previous trials, risperidone showed a lower risk of EPS than the high potency APDs such as haloperidol (RR 0.26; 95% CI 0.10-0.64), we did not observe a lower EPS rate than low-potency APDs (risperidone vs thioridazine RR 1.73, 95% CI 0.49-6.13; risperidone vs pipamperone RR 2.50, 95% CI 0.78-8.04). CONCLUSIONS The reduced EPS rates observed when comparing risperidone with high-potency antipsychotics such as haloperidol may not apply to comparisons with low-potency drugs.