Twenty patients with asymmetric Parkinson's disease were studied in a reaction time (RT) experiment in which the performance of the more affected ("bad") hand was compared with performance of the less affected ("good") hand. Simple RT and choice RT were tested in separate blocks, and the benefit afforded by advance information in the simple RT condition (choice RT minus simple RT) served as a measure of motor preparation. RT was longer in the "bad" hand in both the simple RT and choice RT conditions. There was no difference in the effect of advance information between the two hands. It is concluded that slowness in RT movement initiation in Parkinson's disease is not due to a deficiency in motor preparation, and that intact basal ganglia function is not required for this stage of motor programming.