In a case-control study there were significantly fewer smokers among patients with ulcerative colitis both at disease onset and at interview, the relative risk compared with non-smokers being 0.33 and 0.12, respectively. Among female patients with Crohn's disease there were significantly more smokers, the relative risk being 2.70 and 2.33, respectively. Of the ex-smokers with ulcerative colitis, two-thirds became ill after they stopped smoking, and most of these during the first years after stopping. Neither in ulcerative colitis nor in Crohn's disease could any relation be found between the localization of disease and smoking habits at the time of diagnosis. The findings in the present study support the hypothesis that smoking may influence the course of inflammatory bowel disease.