The purpose of this investigation was to clarify abnormalities in the stride length-cadence relation in gait hypokinesia in Parkinson's disease (PD). A second aim was to investigate the effect of levodopa medication on the foot-step pattern. In the first experiment, 20 subjects with idiopathic PD and 20 age-, sex-, and height-matched controls performed a series of 10 m walking trials at cadence rates ranging from 40 steps/min to 180 steps/min. Cadence rates were set by an electronic metronome, and gait patterns were measured by using a footswitch stride-analyzer system. By instructing subjects to concentrate on walking in time to the metronome beat, the baseline stride length could be monitored for a range of velocities with the compensatory effects of cadence removed. Linear-regression analysis revealed that the mean slope for the regression of stride length against cadence was not different from normal in PD, although there was a statistically significant difference in mean intercept between the PD group (0.25) and the control group (0.59); [t (19) = -4.76; p = 0.0001]. The second experiment evaluated the effects of levodopa on stride-length regulation in 10 subjects with idiopathic PD on average 45 min before and after the first morning dose was administered. There was a statistically significant increase in stride length for all cadence rates from premedication to postmedication phases and the maximal stride length was achieved at higher cadence rates after medication. The slope of the regression of stride length against cadence did not alter according to medication status, although the mean intercept was significantly lower before levodopa (-0.06) compared with after levodopa (0.27); [t (9) = -3.83; p = 0.004]. These results suggest that defective scaling of stride length underlies gait disturbance in PD.