We examined whether people with Parkinson's disease (PD) have a central amplitude regulation disorder using three-dimensional (3-D) gait analyses to compare the effects of medication and attentional strategies on gait in 12 PD subjects and 12 matched comparison subjects. Subjects with PD first performed several 10-m gait trials at preferred speed while off levodopa. They then walked at preferred speed on levodopa; off levodopa with cues; and on levodopa with cues. Control subjects walked at preferred speed and then with visual cues to match their stride length to PD values. As well as spatiotemporal footstep data, pelvic and lower limb kinematic profiles and angle-angle diagrams were produced for sagittal, coronal, and transverse plane movements using a 3-D motion analysis system. In people with PD, decreased step length was accompanied by reduced movement amplitude across all lower limb joints, in all movement planes. When control subjects were required to walk with short steps matched to the size of PD comparisons, they displayed a similar multijoint reduction in amplitude. For PD subjects, both levodopa and visual cues increased movement amplitude across all lower limb joints. Amplitude increased further when levodopa and visual cues were combined, resulting in normalization of step length. This finding suggested that reduced step length is due to a mismatch between cortically selected movement amplitude and basal ganglia maintenance mechanisms. Levodopa and cues normalized amplitude across all joints by altering motor set and bypassing defective basal ganglia mechanisms.