OBJECTIVE: To systematically review the available evidence on the effectiveness of nonpharmacological rehabilitation interventions for people with Parkinson's disease, and identify future research needs. DESIGN: Electronic searches of four databases (CINAHL, Cochrane Library, MEDLINE, PsychLIT) 1980-2002; examination of reference lists of relevant papers. Controlled trials and observational studies were included. Data extraction and quality assessment of papers by two independent reviewers. A narrative review. SETTING: Rehabilitation interventions delivered either in subjects' own homes, or in clinical settings as outpatients. SUBJECTS: Community-living adults with Parkinson's disease. INTERVENTIONS: Physiotherapy, occupational therapy, speech and language therapy, psychological counselling and support, and education. MAIN MEASURES: A range of outcomes: mobility, functional status, speech, swallowing, psychological well-being, as determined by the studies included in the review. RESULTS: Forty-four different studies (reported in 51 papers) were included (25 physiotherapy, 4 occupational therapy, 10 speech and language therapy, 3 psychological counselling, 1 educational, 1 multidisciplinary). All studies, except one, reported improvements on at least one outcome measure. CONCLUSIONS: Findings may reflect publication bias, but suggest interventions can affect patients' lives for the better in a variety of ways. It is difficult to interpret the clinical importance of statistically significant improvements reported in most studies. There is a need for methodologically more robust research with meaningful follow-up periods, designed in a manner that separates specific and nonspecific effects. Cost-effectiveness evidence is required to provide clear guidance on service extensions.