In a case-control study 280 patients with inflammatory bowel disease and matched community controls taken from general practitioner lists were questioned about their smoking habits. Whether assessed by current or previous habits, patients with Crohn's disease were more likely to be or have been smokers than their matched controls and the association appeared strongest at disease onset (matched relative risk 4.1, p less than 0.001). In contrast, patients with ulcerative colitis assessed in the same way were less likely to be smokers compared with their matched controls, with the association being equally strong at disease onset and currently (matched relative risk 0.17, p less than 0.001). Among patients with ulcerative colitis who were exsmokers, 76% reported stopping smoking before disease onset compared with 13% of exsmokers with Crohn's disease. Contrary to earlier reports, there were only small and nonsignificant increases in the risk of Crohn's disease or ulcerative colitis in exsmokers. The strength, temporal precedence, and contrasting nature of the associations are evidence that smoking has an important etiologic role in inflammatory bowel disease. Smoking appears to be a factor in determining whether Crohn's disease or ulcerative colitis develops in predisposed individuals.