Mild cognitive impairment, especially executive dysfunction might occur early in the course of Parkinson's disease. Cognitive training is thought to improve cognitive performance. However, transfer of improvements achieved in paper and pencil tests into daily life has been difficult. The aim of the current study was to investigate whether a multimodal cognitive rehabilitation programme including physical exercises might be more successful than cognitive training programmes without motor training. 240 PD-patients were included in the study and randomly allocated to three treatment arms, group A cognitive training, group B cognitive training and transfer training and group C cognitive training, transfer training and psychomotor and endurance training. The primary outcome measure was the ADAS-Cog. The secondary outcome measure was the SCOPA-Cog. Training was conducted for 4 weeks on a rehabilitation unit, followed by 6 months training at home. Caregivers received an education programme. The combination of cognitive training using paper and pencil and the computer, transfer training and physical training seems to have the greatest effect on cognitive function. Thus, patients of group C showed the greatest improvement on the ADAS-Cog and SCOPA-COG and were more likely to continue with the training programme after the study.