We devised a single-blind study to assess the role of providing external sensory cues in the rehabilitation of patients with idiopathic Parkinson's disease (PD). Twenty stable, nondemented patients with PD entered a 6-week rehabilitation program and were randomly assigned to two balanced protocols which were differentiated by the use of external sensory cues ("non-cued" vs "cued"). Patients were evaluated by a neurologist, who was blind to group membership, with the Unified Parkinson's Disease Rating Scale (UPDRS) at baseline, end of treatment, and after 6 weeks. Patient groups were comparable for age, disease duration, and severity. A significant reduction of UPDRS scores (activities of daily living and motor sections) was present after the rehabilitation phase in both groups. However, at follow up, while this clinical improvement had largely faded in the "non-cued" group, mean UPDRS scores of the "cued" group were still significantly lower than baseline values. The incorporation of external sensory cues in the rehabilitation protocol can extend the short-term benefit of physical therapy in moderately disabled patients with PD, possibly as a result of the learning of new motor strategies. "Cued" physical therapy for PD should be targeted to compensate for the defective physiological mechanisms.