BACKGROUND Strength training or aerobic exercise programmes might maximise muscle and cardiorespiratory function and prevent additional disuse atrophy in patients with muscle disease. However, over-exerting might cause more rapid disease progression. OBJECTIVE To examine the efficacy and safety of strength training and aerobic exercise training in patients with muscle diseases. METHODS We searched the Cochrane Neuromuscular Disease Group register (October 2002 and May 2004), the Cochrane Collaboration Rehabilitation and Related Therapies Field register (October 2002), MEDLINE (January 1966 to December 2002), EMBASE (January 1973 to October 2002), and CINAHL (January 1982 to August 2002) for randomised trials. We reviewed the bibliographies of trials identified and reviews covering the subject. METHODS Randomised or quasi-randomised controlled trials comparing strength training and/or aerobic exercise programmes lasting at least 10 weeks. Types of outcome measures: FOR STRENGTH TRAINING. Primary: static or dynamic muscle strength. Secondary: muscle strength (endurance or fatigue), functional assessments, quality of life, muscle membrane permeability, pain, and fatigue. FOR AEROBIC EXERCISE TRAINING. Primary: aerobic capacity expressed as work capacity. Secondary: aerobic capacity (oxygen consumption, parameters of cardiac or respiratory function), functional assessments, quality of life, muscle membrane permeability, pain, and fatigue. METHODS Two reviewers independently assessed trial quality and extracted the data. RESULTS We identified two randomised trials fulfilling all inclusion criteria. The first trial compared the effect of strength training versus no training in 36 patients with myotonic dystrophy. The other trial compared strength training versus no training combined with albuterol or placebo in 65 patients with facioscapulohumeral muscular dystrophy. Methodological quality and training programmes were graded adequate. In the myotonic dystrophy trial there were no significant differences between training and non-training groups for the primary outcome measure. In the facioscapulohumeral muscular dystrophy trial static muscle strength did not show significant differences between training and non-training groups. Only a +1.2 kg difference (95% confidence interval 0.2 to 2.1) in dynamic strength of elbow flexors in favour of the training group, reached statistical significance. For both trials there were no significant differences between groups for most of the secondary outcome measures, including those covering adverse effects. CONCLUSIONS In myotonic dystrophy and facioscapulohumeral muscular dystrophy moderate-intensity strength training appears not to do harm but there is insufficient evidence to establish that it offers benefit. Limitations in the design of studies in other muscle diseases prevent general conclusions in these disorders.