A randomized clinical trial was conducted to determine whether colonoscopy is useful in deciding how long to maintain steroid treatment in attacks of Crohn's disease involving the colon. One hundred forty-seven patients with acute attacks of colonic or ileocolonic Crohn's disease were treated by oral prednisolone, 1 mg.kg-1.day-1; 136 achieved clinical remission, but 96 of them still had active endoscopic lesions and were randomized either to immediate start of steroid tapering (group A; n = 46) or to continued prednisolone treatment at the same dosage for 5 more weeks before steroid tapering was begun (group B; n = 50). In the remaining 40 patients (already in endoscopic remission, group C), steroid tapering was begun immediately. After prednisolone discontinuation, patients were followed up for 18 months or until clinical relapse. Prolongation of prednisolone therapy significantly improved the endoscopic scores in group B (30% of endoscopic remission). The frequency of successful steroid weaning was almost identical in groups A and B (82% and 80%, respectively), as was the actuarially calculated relapse clinical rate after steroid withdrawal (P = 0.22). No factor predictive of clinical relapse could be found. The clinical course of patients in group C was similar to that of those in groups A and B. Overall, only 22% of the 147 patients were still in clinical remission and off steroids 18 months after prednisolone discontinuation, outlining the need for maintenance therapy. In conclusion, for patients who have achieved clinical remission, adjustment of steroid treatment duration on the basis of endoscopy results is of no benefit, and the endoscopic aspect has no prognostic value; thus, it appears unnecessary to repeat colonoscopy in such patients before steroid tapering is begun.