In this study, the extent to which bradykinesia in patients with idiopathic Parkinson's disease can be influenced by practice and by specific training strategies was investigated. Fifteen patients with Parkinson's disease tested after withdrawal of anti-Parkinson medication, and 15 matched control subjects, practised a ballistic aiming task. Performance was tested before, during and after training and again 1 h later. The Parkinson's disease patients and control subjects were randomly assigned to one of two training schedules, practising with or without rhythmic auditory cues. At baseline, the Parkinson's disease patients showed longer movement times, with a marked decrease in maximum acceleration and deceleration in the initial open-loop phase compared with those of the control subjects. With training, they were able to make significant improvement in the speed of aimed movements, particularly in the early movement phase, without any deterioration in accuracy. These effects transferred to an untrained limb and were at least partially maintained after a 1-h delay. While patients remained impaired relative to control subjects at all phases of training and follow-up, the patients' performance at the end of training did not differ significantly from the control subjects' baseline function. Contrary to expectation, rhythmic auditory cues did not enhance improvement in the speed of aimed movements in either patients or control subjects. If anything, less improvement was shown in the cued groups, although there were suggestions that the aiming skill was retained better over the delay period. The results demonstrate preserved abilities to improve speed of single ballistic aiming movements in Parkinson's disease patients and the possibility of reducing bradykinesia by training.