We studied 45 non-demented patients with Parkinson's disease (PD), 12 with progressive supranuclear palsy (PSP), 10 with multiple system atrophy (MSA) and 12 with neuroleptic-induced parkinsonism (NIP) for the presence of apraxia. Our aim was to determine whether a standard comprehensive assessment of different praxic functions would demonstrate specific types of errors not attributable to bradykinesia, rigidity, tremor or any other abnormal elementary motor deficit. PD patients on chronic levodopa treatment were examined in the 'on' and 'off' (treatment) states. Based on apraxia assessment scores, bilateral ideomotor apraxia for transitive movements was found in eight (75%) and 12 (27%) of PSP and PD patients, respectively. Ideomotor apraxia was mainly characterized by spatial errors (i.e., external and internal configuration, body-part-as-object and trajectory). Four PSP but no PD patients exhibited ideomotor apraxia for intransitive movements. PSP as well as PD patients with ideomotor apraxia also had difficulties in imitating hand and finger postures, but none of them failed on pantomime comprehension and pantomime recognition/discrimination. Some PSP patients exhibited, in addition, a limbkinetic type of apraxia and a minority of them displayed deficits on tasks involving multiple steps. Neither MSA nor NIP patients showed any disturbance of praxic functions. There were no differences in age, disease duration, Mini Mental State Examination (MMSE), Unified Parkinson's disease Rating Scale and Hoehn-Yahr scores between apraxic and non-apraxic PD patients, and ideomotor apraxia scores were similar in the 'on' and 'off' states. A correlation was found between ideomotor apraxia scores in PD patients and deficits in frontal lobe-related neuropsychological tasks such as the Tower of Hanoi, verbal fluency and the Trail Making Test. Furthermore, PD patients with apraxia showed higher Hamilton depression scores than non-apraxic PD patients. In PSP patients, ideomotor apraxia scores correlated significantly with cognitive deficit as measured with MMSE. The presence or absence of cortical involvement, and its severity and distribution might determine the presence and type of apraxia in PD and PSP. Apraxia in these conditions would therefore reflect combined cortico-striatal dysfunction.