We evaluated the efficacy of physiotherapy, occupational therapy, and speech and language therapy in Parkinson's disease by synthesizing six Cochrane systematic reviews. All randomised, controlled trials examining the efficacy of a paramedical therapy versus control intervention and all those comparing the efficacy of two forms of active therapy in Parkinson's disease were included. Trials were identified by searching biomedical databases, reference lists, hand searching, and contacting investigators. The main outcome measures were quality of life, speech intelligibility, activities of daily living, and individual measures of motor and speech impairment. We identified 16 physiotherapy randomised controlled trials (399 patients), two occupational therapy trials (84 patients), and five speech and language therapy for dysarthria trials (154 patients). None of these studies examined nonpharmacological swallowing therapy for dysphagia. We were unable to perform meta-analysis of the results because the trials used heterogeneous therapy methods and outcome measures. The trials also had marked methodological flaws that could have introduced bias. In summary, we failed to find conclusive evidence of benefit for any form of paramedical therapy sufficient to recommend them in routine clinical practice. However, this lack of evidence is not proof of a lack of effect. Further large pragmatic randomised controlled trials are required to determine the effectiveness of paramedical therapies in Parkinson's disease. © 2002 Movement Disorder Society