We have reviewed the clinical and pathological diagnoses of 143 cases of parkinsonism seen by neurologists associated with the movement disorders service at The National Hospital for Neurology and Neurosurgery in London who came to neuropathological examination at the United Kingdom Parkinson's Disease Society Brain Research Centre, over a 10-year period between 1990 and the end of 1999. Seventy-three (47 male, 26 female) cases were diagnosed as having idiopathic Parkinson's disease (IPD) and 70 (42 male, 28 female) as having another parkinsonian syndrome. The positive predictive value of the clinical diagnosis for the whole group was 85.3%, with 122 cases correctly clinically diagnosed. The positive predictive value of the clinical diagnosis of IPD was extremely high, at 98.6% (72 out of 73), while for the other parkinsonian syndromes it was 71.4% (50 out of 70). The positive predictive values of a clinical diagnosis of multiple system atrophy (MSA) and progressive supranuclear palsy (PSP) were 85.7 (30 out of 35) and 80% (16 out of 20), respectively. The sensitivity for IPD was 91.1%, due to seven false-negative cases, with 72 of the 79 pathologically established cases being diagnosed in life. For MSA, the sensitivity was 88.2% (30 out of 34), and for PSP it was 84.2% (16 out of 19). The diagnostic accuracy for IPD, MSA and PSP was higher than most previous prospective clinicopathological series and studies using the retrospective application of clinical diagnostic criteria. The seven false-negative cases of IPD suggest a broader clinical picture of disease than previously thought acceptable. This study implies that neurologists with particular expertise in the field of movement disorders may be using a method of pattern recognition for diagnosis which goes beyond that inherent in any formal set of diagnostic criteria.