INTRODUCTION: Antiparkinsonian drugs can induce behavioural disturbances, which should be treated by first reducing antiparkinsonian drugs and/or starting a benzodiazepine. If this approach fails, then antipsychotics can be considered. The aim of tis study was to determine how often antiparkinsonian drugs are decreased and benzodiazepines are started in levodopa users before start of an antipsychotic drug. METHODS: Data came from the PHARMO database, which includes drug dispensing records for allresidents of six Dutch cities. All patients were selected who had started antipsychotic drug therapy at least 360 days later than, and 180 days before the earliest and the latest dispensing date for levodopa in PHARMO respectively. For each patient who started an antipsychotic drug we randomly selected 2 matched controls who used levodopa for at least 540 days and who did not start an antipsychotic. We determined how many persons had reduced antiparkinsonian drug treatment (i.e., dosage decrement or discontinuation of an antiparkinsonian agent) or had started a benzodiazepine in the 180 days before the start of the antipsychotic or before a randomly chosen index date in the controls. RESULTS: We identified 40 antipsychotic starters and 64 matched controls. The prescribed daily dose of antiparkinsonian drug treatment was reduced in 14 antipsychotic starters (37%) and in 17 controls prior to the index date (27%) (relative risk 1.27, 95% confidence interval 0.77-2.08). Of these, 2 antipsychotic starters(5%) and 5 controls (8%) had started with a benzodiazepine (relative risk 0.73; 0.22-2.96). A further 34 antipsychotic starters (85%) and 49 controls (77%) returned later than expected to their community pharmacy before the index date (relative risk 1.43; 0.70-2.96). CONCLUSION: Our study demonstrates that the advice to reduce the levodopa dose or to start with a benzodiazepine when behavioural disturbances occur during levodopa treatment prior to the start of an antipsychotic, is not followed in dailypractice.