Elemental diet is as effective as corticosteroids in the treatment of previously untreated Crohn's disease. It is unclear whether a poor nutritional state is a prerequisite for efficacy of elemental diet, whether previously treated patients respond as well, or how duration of remission using elemental diet compares with corticosteroid induced remission. Forty two patients with active Crohn's disease were stratified for nutritional state and randomised to receive Vivonex TEN 2.1 l/day for four weeks, or 0.75 mg prednisolone/kg/day for two weeks and subsequent reducing doses. Nine of 22 (41%) patients assigned to nutritional treatment were intolerant of the diet. Thirty patients completed four weeks treatment. Disease activity decreased on elemental diet from mean (SEM) 4.8 (0.9) to 1.7 (0.6), p < 0.05, and on prednisolone from 5.3 (0.5) to 1.9 (0.6), p < 0.05. For each treatment, nourished and malnourished patients responded similarly. Patients with longstanding disease responded as well as newly diagnosed patients. The probability of maintaining remission at six months was 0.67 after prednisolone, 0.28 after elemental diet, and at one year was 0.35 after prednisolone and 0.09 after elemental diet, p < 0.05. When tolerated, elemental diet is as effective in the short term as prednisolone in newly and previously diagnosed Crohn's disease, and its benefit is independent of nutritional state. The subsequent relapse rate after elemental diet induced remission, however, is greater than after treatment with prednisolone.