BACKGROUND & AIMS: Barrett's esophagus (BE) is the precursor to esophageal adenocarcinoma; the incidence of this cancer is rapidly increasing in Western populations. However, few population-based studies of BE have been conducted, so little is known about potentially modifiable causes of this disease. METHODS: The study included patients with newly diagnosed BE, confirmed by histology and categorized as simple BE (without dysplasia, n = 285) or dysplastic BE (with dysplasia, n = 108). We recruited 2 separate control groups: endoscopy patients with acute inflammatory changes (inflammation controls, n = 313) and population controls sampled from a population register (n = 644). Data were collected through standardized questionnaires and telephone interviews. We fit logistic regression models to calculate odds ratios (ORs) for BE associated with salient exposures by using each set of controls. RESULTS: Relative to never smokers, risks of simple BE were significantly higher among former smokers (OR, 2.39; 95% confidence interval, 1.59-3.60) and current smokers (OR, 2.41; 95% confidence interval, 1.39-4.17), compared with population controls. Smoking conferred more than a 4-fold increase in risk for dysplastic BE; this increase in risk remained long after individuals quit smoking. We found no conclusive association between BE and passive smoking and no evidence of independent associations between body mass index (BMI) and simple BE or dysplastic BE, after adjusting for reflux and other factors. Analyses with inflammation controls produced qualitatively similar risk estimates for smoking and BMI to those obtained for population controls, but they were markedly attenuated for reflux, as expected. CONCLUSIONS: Current and past smoking significantly increases risk for BE, but BMI does not, after adjustment for the effect of reflux.