Background. Patients with Parkinson's disease (PD) are encouraged to stay active to maintain their mobility. Ambulatory activity monitoring (AM) provides an objective way to determine type and amount of gait-related daily activities. Objective. To investigate the effects of a home cueing training program on functional walking activity in PD. Methods. In a single-blind, randomized crossover trial, PD patients allocated to early intervention received cueing training for 3 weeks, whereas the late intervention group received training in the following 3 weeks. Training was applied at home, using a prototype cueing device. AM was applied at baseline, 3, 6, and 12 weeks in the patient's home, to record body movements. Postures and motions were classified as percentage of total time spent on (a) static activity, further specified as % sitting and % standing, and (b) % dynamic activity, further specified as % walking, % walking periods exceeding 5 seconds (W>5s) and 10 seconds (W> 1 0s). Random coefficient analysis was applied. Results. A total of 153 patients participated in this trial. Significant improvements were found for dynamic activity (b = 4.46; P < .01), static activity (b = -3.34; P < .01), walking (b = 4.23; P < .01), W>5s (b = 2.63; P < .05), and W> 1 0s (b = 2.90; P < .01). All intervention effects declined significantly at 6 weeks follow-up. Conclusion. Cueing training in PD patients' own home significantly improves the amount of walking as recorded by AM. Treatment effects reduced after the intervention period, pointing to the need for permanent cueing devices and follow-up cueing training.