To summarize and critically appraise the available evidence on exercise therapy and other types of physical therapies for patients with neuromuscular diseases (NMD). Cochrane Central Register of Controlled Trials and Cochrane Database of Systematic Reviews, Medline, CINAHL, EMBASE (Rehabilitation and Physical Medicine), and reference lists of reviews and articles. Randomized clinical trials (RCTs), controlled clinical trials (CCTs), and other designs were included. Study participants had to have any of the following types of NMD: motoneuron diseases, disorders of the motor nerve roots or peripheral nerves, neuromuscular transmission disorders, or muscle diseases. All types of exercise therapy and other physical therapy modalities were included. Outcome measures had to be at the level of body functions, activities, or participation according to the definitions of the International Classification of Functioning, Disability and Health (ICF). Two reviewers independently decided on inclusion or exclusion of articles and rated the methodologic quality of the studies included. All RCTs, CCTs, and other designs only if of sufficient methodologic quality were included in a best evidence synthesis. A level of evidence was attributed for each subgroup of NMD and each type of intervention. Initially 58 studies were included: 12 RCTs, 5 CCTs, and 41 other designs. After methodologic assessment, 19 other designs were excluded from further analysis. There is level II evidence ("likely to be effective") for strengthening exercises in combination with aerobic exercises for patients with muscle disorders. Level III evidence ("indications of effectiveness") was found for aerobic exercises in patients with muscle disorders and for the combination of muscle strengthening and aerobic exercises in a heterogeneous group of muscle disorders. Finally, there is level III evidence for breathing exercises for patients with myasthenia gravis and for patients with myotonic muscular dystrophy. Adverse effects of exercise therapy were negligible. The available evidence is limited, but relevant for clinicians. Future studies should be preferably multicentered, and use an international classification of the variables of exercise therapy and an ICF core set for NMD in order to improve comparability of results.