The effects of different types of advance information (warning signal, temporal cue, movement parameter cue), presented at different intervals before an imperative stimulus on reaction time (RT) were examined. Sixteen patients with Parkinson's disease and 16 age-matched normal controls performed a simple reaction time (SRT) task, an uncued, two partially cued (cueing of direction of movement or hand) and a fully cued visual four choice reaction time (CRT) tasks. An S1-S2 paradigm was used, where a warning signal/precue (S1) preceded the imperative signal (S2) by 0 (uncued and unwarned conditions), 200 ms, 800 ms, 1600 ms or 3200 ms. When the unwarned SRT and uncued CRT conditions were compared, the patients were slower than the controls in both conditions, with the group differences being larger for the CRT than the SRT task. Both patients and controls benefited from a warning signal presented before the imperative stimulus (S2), especially when it occurred 200 ms prior to S2. Advance information about the parameters of the required response was used by both patients and controls to preprogramme responses before the onset of S2. In both groups, RTs were generally faster for longer S1-S2 intervals. With an S1-S2 interval of 3200 ms, advance information about two movement parameters was 'fully' used by the patients to preprogramme responses such that their fully cued CRT was reduced to the level of SRT. In the controls, this process was complete by 800 ms. The most parsimonious explanation of the pattern of results across the SRT, fully cued CRT and uncued CRT conditions would be in terms of a slowness in response initiation in Parkinson's disease, which is a stage of processing common to all RT conditions. Deficits at one or more of the stages of processing unique to CRT are, however, necessary to explain the differentially greater slowness of uncued CRT in Parkinson's disease.