Previous studies suggested that methotrexate has beneficial effects in patients with Crohn's disease. We report our experience with this agent in patients with chronic active Crohn's disease who previously failed to improve with conventional treatment, including azathioprine in most cases. Between June 1988 and June 1992, 39 patients with refractory Crohn's disease were treated with methotrexate. In patients with active disease, clinical remission was defined by a Harvey-Bradshaw index of less than 4. For patients also taking corticosteroids, the dates of remission and complete steroid withdrawal were recorded. For patients who achieved clinical remission, and those in clinical remission when methotrexate was started, the relapse rate on methotrexate therapy was noted. In the 37 patients with active disease at methotrexate initiation, the probability of remission was 72% at 3 months. The probability of remission and steroid withdrawal was 42% at 12 months. In patients on clinical remission, the probability of relapse on methotrexate was 58% at 12 months. Twenty-two patients experienced side-effects, but these only warranted methotrexate discontinuation in four cases. Methotrexate appears effective in most patients with refractory Crohn's disease and its short-term toxicity is acceptable, but the long-term benefit seems more limited.