Many authorities have drawn attention to the difficulties in clinically distinguishing Parkinson's disease (PD) from other parkinsonian syndromes. We assessed the clinical features of 100 patients diagnosed prospectively by a group of consultant neurologists as having idiopathic PD according to their pathologic findings. Seventy-six percent of these cases were confirmed to have PD. By using selected criteria (asymmetrical onset, no atypical features, and no possible etiology for another parkinsonian syndrome) the proportion of true PD cases identified was increased to 93%, but 32% of pathologically confirmed cases were rejected on this basis. These observations suggest that studies based on consultant diagnosis of PD, using standard diagnostic criteria, will include cases other than PD, thus distorting results from clinical trials and epidemiologic studies. The strict use of additional criteria can reduce misdiagnosis but at the cost of excluding genuine PD cases.