We screened a random sample of 2449 people aged 65 years and over for undiagnosed parkinsonism, using a postal screening questionnaire followed by clinical neurological assessment. Amongst the 1556 (63.5%) patients who responded, four patients with previously undiagnosed parkinsonism were identified, suggesting a prevalence of 257 per 100,000 (95% CI 70, 658) in this age-group. Although only small, the numbers were sufficient to significantly increase the incidence of parkinsonism in an incidence study. Two simple screening questions achieved a high sensitivity for newly diagnosed parkinsonism of 95%, but a low specificity of 28%.