The aim of our study was to analyse decision making in early-onset Parkinson's disease (PD) patients performing the Iowa Gambling Task (IGT). We compared 19 patients with early-onset PD (<45 years) on dopaminergic medication (no evidence of depression, dementia, executive dysfunction according to the Tower of London test and the Stroop test, or pathological gambling) with 20 age-matched controls. A computer version of the IGT was employed. The PD patients achieved slightly lower IGT scores than the control group. A detailed analysis based on 'shift frequencies' between the individual decks showed that the patients tended to change their preferences for the decks more frequently, with a higher preference for the 'disadvantageous' deck B. Control subjects seemed to develop a more effective strategy. These differences could be caused by the poorer ability of the patients to develop any strategy at all. We observed changes in decision making during IGT performance in patients with early-onset PD, although they had no executive dysfunction as measured by established neuropsychological tests. The more detailed analysis employed in the present study could lead to a more accurate study of IGT performance and application of IGT in clinical practice.