Background: The nature of cognitive performance in subjects with Parkinson disease (PD) without dementia is controversial, perhaps because of failure to exclude subjects with unrecognized very mild dementia. Objective: To compare cognitive and motor functioning in well-characterized subjects with PD without overt dementia with healthy elderly control subjects. Design: Subjects' conditions were evaluated clinically and psychometrically at entry into a longitudinal study of cognitive and motor performance in elderly subjects. Measures included a global dementia staging scale, the Washington University Clinical Dementia Rating; psychometric tests, including Logical Memory, Digit Span, Associate Learning, Information, Block Design, Digit Symbol, Trailmaking A, Crossing-off, Boston Naming Test, and Word Fluency; and motor measures, including finger tapping, gait velocity, reaction time, and movement time. Settings A university-based research facility. Subjects: There were 3 groups of subjects: healthy elderly control subjects (n = 43), subjects with PD without dementia (n = 58), and subjects with PD with questionable dementia (n = 22), each evaluated at time of entry. Results: As expected, both PD groups were impaired on motor measures (gait velocity, finger tapping, and movement time) compared with the healthy elderly control group. Neither PD group showed slowing in reaction time. The subjects with PD with questionable dementia were more impaired on Logical Memory, Block Design, Digit Symbol, and Trailmaking. A compared with the subjects with PD without dementia. Although free of clinically evident cognitive dysfunction (Clinical Dementia Rating score, 0), the PD group without dementia was impaired with respect to the healthy elderly control group on all measures from the psychometric assessment except Digit Span, Associate Learning, and Word Fluency. Conclusions: The PD group without dementia showed global cognitive impairments in comparison with the healthy elderly control group, possibly because the healthy elderly control subjects represented idealized aging. Although the deficits were of small magnitude, this finding suggests that PD may predispose to subclinical cognitive impairment. Longitudinal follow-up is required to determine whether subjects with PD destined to develop overt dementia can be distinguished from those who do not.