Patients with Parkinson's disease (PD) walk with a shortened stride length and high stride-to-stride variability, a measure associated with fall risk. Rhythmic auditory stimulation (RAS) improves stride length but the effects on stride-to-stride variability, a marker of fall risk, are unknown. The effects of RAS on stride time variability, swing time variability and spatial-temporal measures were examined during 100-m walks with the RAS beat set to 100 and 110% of each subject's usual cadence in 29 patients with idiopathic PD and 26 healthy age-matched controls. Carryover effects were also evaluated. During usual walking, variability was significantly higher (worse) in the patients with PD compared with the controls (P < 0.01). For the patients with PD, RAS at 100% improved gait speed, stride length and swing time (P < 0.02) but did not significantly affect variability. With RAS at 110%, reductions in variability were also observed (P < 0.03) and these effects persisted 2 and 15 min later. In the control subjects, the positive effects of RAS were not observed. For example, RAS increased stride time variability at 100 and 110%. These results demonstrate that RAS enables more automatic movement and reduces stride-to-stride variability in patients with PD. Further, these improvements are not simply a by-product of changes in speed or stride length. After walking with RAS, there also appears to be a carryover effect that supports the possibility of motor plasticity in the networks controlling rhythmicity in PD and the potential for using RAS as an intervention to improve mobility and reduce fall risk.