A group of ten patients with idiopathic Parkinson's disease (PD) was tested on a series of automated tests of learning, memory, planning and attention whilst either on or off L-dopa medication. Controlled withdrawal of L-dopa interfered with aspects of performance on three of the tests that had previously been shown to be sensitive to frontal lobe dysfunction; a spatial working memory task, the Tower of London planning test, and a visual discrimination paradigm that also included intra- and extra-dimensional shift tests of selective attention. More specifically, errors were increased in the spatial working memory test, and both the accuracy and latency of thinking were impaired. Thinking time was significantly slowed following L-dopa withdrawal, even though the possible contaminating effects on motor slowing were fully controlled by a yoked control procedure. Nine out of ten patients reached a further stage of the visual discrimination, set-shifting paradigm when on, rather than off, L-dopa medication. Spatial span was also impaired off medication, but there were no effects of L-dopa withdrawal on tests of pattern and spatial recognition memory, simultaneous and delayed matching to sample or visuospatial conditional associative learning. Comparisons with a large control group confirmed previous findings that PD is associated with deficits on the majority of these tests. The results are discussed in terms of the fronto-striatal, dopamine dependent nature of some of the cognitive deficits found in PD, but the apparent dopamine-independent nature of deficits in other aspects of cognitive functioning, notably in tests of visual recognition memory and associative learning.