We randomly assigned 39 patients with steroid-dependent generalized myasthenia gravis to treatment with cyclosporine (5 mg/kg per body weight in divided doses) or placebo. Duration of treatment was 6 months. Patients were evaluated monthly. Primary measures of efficacy were quantified strength testing, antihuman acetylcholine receptor antibody titer, and dosage of corticosteroid medication. At the end of the study, patients in the cyclosporine group had significantly greater improvement in strength (p = 0.004) and a reduction in antireceptor antibody titer (p = 0.01). Percentage reduction of steroid medication was greater in the cyclosporine group, although the difference was not statistically significant (p = 0.12). There were no treatment failures, and there was one drug failure in the cyclosporine group. In the placebo group, there were three treatment failures. No significant nephrotoxicity was noted at this dosage during the first 6 months. During the subsequent 18 months of open-label therapy, continued reduction in steroid dosage occurred. Cumulative side effects, however, caused 35% of patients to discontinue the medication; 10% did so secondary to slowly progressive nephrotoxicity.