A positive association between tobacco and colorectal adenomas has been suggested. Smoking is, however, also associated with 'poor' dietary habits, which in turn may be related to risk of adenomas. It is therefore of interest to study the relationship between smoking, diet and risk of colorectal adenomas in follow-up studies. We compared 87 adenoma cases to 35 'hospital' and 35 healthy controls (all controls were age- and sex-matched and proven to be free of adenomas). Smoking data were collected by an interview and a self-administrated questionnaire with a time interval of at least one month. After 3 years of follow-up, all polyps were removed. Our data indicate that smoking is associated with adenoma prevalence, but not necessarily with size, multiplicity, growth or recurrence of adenomas. Compared to both sets of controls, cases reported to have smoked more than 15 pack-years, or who are current smokers, had a fourfold increased frequency of adenomas (odds ratios 3.6-5.9). Smokers with adenomas had dietary habits that may also be associated with adenomas. The smoking estimates remained largely unchanged even after adjustments for dietary variables in multivariate analysis. This study lends support to the theory of an initiating role of tobacco smoke in neoplasia formation.