In the European Cooperative Crohn's Disease Study patients from 14 centers were included in whom diagnosis was made within 2 years before study entry on the basis of generally accepted radiological, endoscopical and/or histological criteria or a combination of all. Reasons for exclusion were: diagnosis older than 2 years in patients who did not require active treatment, age less than 18 years, duration of symptoms less than 3 months, presence of complications which potentially required emergency surgery. Data on clinical features were obtained in 633 patients, of whom 452 were eligible to participate in the study. In 110 patients randomized to placebo the natural course of Crohn's disease was studied. Patients with ileocolonic involvement were younger than patients with either colonic or small intestinal involvement only. Classic ileitis terminalis was present in 14% of the patients. 49% of the patients had combined involvement of both the small and large intestine. 30% of patients had only small intestinal involvement, and in 21% colonic disease was present. Small intestinal involvement was associated with a significantly lower Crohn's Disease Activity Index (CDAI) than other anatomical locations of the disease. Perianal disease was more often associated with colonic than with small intestinal involvement. 60% of placebo patients with active disease at entry achieved at least a transient remission within the initial 5 months of study. After 2 years, 23% of patients with active disease at entry and 68% of patients with quiescent disease had reached or maintained a remission, respectively. By stepwise multiple linear regression analysis long duration of disease between diagnosis and randomization, normal serum albumin and combined involvement of small intestine and colon were identified as predictors of a more favorable outcome of patients treated with placebo. In contrast, extensive small bowel disease, treatment with steroids and bowel resection prior to study entry correlated with a less favorable outcome. However, by life table analysis outcome of previously untreated and treated patients in the placebo group was similar.