69 patients with steroid-dependent asthma (mean daily prednisolone dose 14.2 [SD 6.1] mg) took part in a randomised, double-blind, placebo-controlled study of 24 weeks' treatment with methotrexate 15 mg weekly. The patients were seen every 4 weeks by the same physician, who reduced the daily prednisolone dose by 2.5 mg if the daily diary card variables and measurements of lung function were unchanged or improved. All other treatment remained unchanged. After 24 weeks of treatment the prednisolone dose had been reduced by a significantly greater proportion in the methotrexate than in the placebo group (50% vs 14%) and the reduction was not sustained after the study treatment stopped. There were substantial abnormalities in liver function tests in 5 of the 38 patients taking methotrexate.