From January 1983 to October 1990, 41 patients with generalised myasthenia gravis were randomly given either prednisone or azathioprine. The main goal was to record the time to the occurrence of the first episode of deterioration. During a mean follow-up of 30 months, 21 patients showed deterioration, 12 in the prednisone group and nine in the azathioprine group (p = 0.40). No difference was observed between the two groups in muscular score and functional grade, assessed at the end of each treatment year, or in tolerance. Treatment failure occurred in 17 patients, 12 in the prednisone group and five in the azathioprine group (p = 0.02); even after adjustment for imbalances in prognostic features, the failure rate remained 2.8 times higher in the prednisone group than in the azathioprine group (p = 0.5). In the patients in whom treatment failed, symptoms were initially more severe than in the others, but the combination of prednisone and azathioprine resulted in clinical improvement, consisting of remission or only minor deficits in half of the patients after two years of treatment. These findings indicate that azathioprine increases treatment response compared with prednisone, although no difference in the duration of improvement was demonstrated. Nevertheless, it appears that the most severe forms of the disease, often resistant to prednisone or azathioprine alone, could benefit from the combination of both drugs.