Studies of reaction time in Parkinson's disease (PD) have suggested a selective deficit in simple reaction time (SRT), compared with choice reaction time (CRT). This finding has been interpreted as a deficit in motor preprogramming but could involve other factors, such as attentional focussing and stimulus predictability. Moreover, not all studies show the same selective deficit, possibly because of differences in patient selection and treatment effects. The neurochemical basis of RT deficits in PD remains unclear. Accordingly, the contribution of cognitive factors to impaired RT was assessed in a large group of PD patients, including early untreated cases, and performance was examined in relation to clinical variables and the effect of treatment in longitudinal study. Motor output was constant in both SRT and CRT tasks. In the SRT task, all stimuli required a response; in the CRT task, subjects were required to respond to only one of the two possible stimuli. Attentional focussing on SRT was examined by variation of the interval between cue and stimulus; effects of stimulus uncertainty were evaluated from a comparison of SRT and CRT; temporal predictability of the stimulus was examined from a comparison of conditions in which the interval between warning signal and imperative stimulus was constant or variable. The PD patients showed similar deficits in SRT and CRT, but normal effects of cue-stimulus interval and temporal predictability. Reaction time correlated with measures of global cognitive capacity and frontal-lobe function, as well as motor disability. Treatment had no effect on SRT or CRT, despite clinical benefit. These findings indicate that RT deficits in PD are not due to impaired attentional focussing or stimulus predictability but are compatible with a deficit in higher-order processes concerned with the orientation of both cognitive and motor responses to a stimulus. These processes are not substantially dopamine-dependent but may be served by non-dopaminergic neurotransmission.