Changes in gait performance in 153 subjects with PD using three rhythmical cues (auditory, visual and somatosensory) were measured during a simple walking task and a dual walking task in the home. Subjects were 'on' medication and were cued at preferred step frequency. Accelerometers recorded gait and walking speed, step amplitude and step frequency were determined from raw data. Data were analysed with SAS using linear regression models. Gait performance during a single task reduced with cues in contrast to a dual task where PD subjects appeared to benefit from rhythmical cues (increased speed and step length). Effects were dependent on cue modality with significant improvements for auditory cues compared to others. A significant short-term carry-over effect of cues reduced 3 weeks later. Cues may reduce attentional demands by facilitating attentional allocation, accounting for differences of cue seen during single and dual task. Furthermore cue modality may influence attentional demand which is an important consideration for rehabilitation.