To assess if smoking is associated with Graves' disease and, if so, to ascertain whether this association persists when controlling for confounding factors. Consecutive entry case-control study with two age- and sex-matched control subjects from two different populations per case patient. University hospital. Five groups were studied: (1) Graves' ophthalmopathy and Graves' hyperthyroidism (n = 100; divided in four subgroups according to the severity of the eye disease); (2) Graves' hyperthyroidism without clinical eye involvement (n = 100); (3) sporadic nontoxic goiter (n = 100); (4) autoimmune hypothyroidism (n = 75); and (5) toxic nodular goiter (n = 75). The study comprised 200 subjects from a hospital-based population, and 200 from a population-based group served as control subjects. Smoking status was determined from a questionnaire at the time of onset of the disease to exclude any effect of the disease itself on smoking. Smoking greatly increased the risk for Graves' ophthalmopathy (odds ratio, 7.7; 95% confidence interval, 4.3 to 13.7), but patients with Graves' hyperthyroidism alone were also more often smokers than control subjects (odds ratio, 1.9; 95% confidence interval, 1.1 to 3.2). Smoking was not associated with the other thyroid diseases studied. Essentially similar results were obtained after adjustment for differences in education between case patients and control subjects. Among the patients with Graves' ophthalmopathy, smokers had more severe eye disease than nonsmokers, but no association was found between the number of cigarettes smoked per day or the duration of smoking and the severity of the ophthalmopathy. However, there was a significant increase in the odds ratios in patients with more severe eye disease. Smoking is associated with Graves' disease, and it especially increases the risk for the development of more severe ophthalmopathy. Thus, smoking appears to be one of the multiple factors inducing Graves' disease in genetically predisposed individuals.