Groups of patients with idiopathic Parkinson's disease, either medicated or unmedicated, were compared with matched groups of normal controls on a computerized battery previously shown to be sensitive to frontal lobe dysfunction, including tests of planning, spatial working memory and attentional set-shifting. In a series of problems based on the 'Tower of London' test, medicated patients with Parkinson's disease were shown to be impaired in the amount of time spent thinking about (planning) the solution to each problem. Additionally, an impairment in terms of the accuracy of the solution produced on this test was only evident in those patients with more severe clinical symptoms and was accompanied by deficits in an associated test of spatial short-term memory. Medicated patients with both mild and severe clinical symptoms were also impaired on a related test of spatial working memory. In contrast, a group of patients who were unmedicated and 'early in the course' of the disease were unimpaired in all three of these tests. However, all three Parkinson's disease groups were impaired in the test of attentional set-shifting ability, although unimpaired in a test of pattern recognition which is insensitive to frontal lobe damage. These data are compared with those previously published from a group of young neurosurgical patients with localized excisions of the frontal lobes and are discussed in terms of the specific nature of the cognitive deficit at different stages of Parkinson's disease.