Obesity has been associated with an increased risk of cancer at a number of sites. A notable exception appears to be lung cancer, for which several studies suggest a modest inverse association. However, cigarette smoking is directly associated with lung cancer and inversely associated with body mass index. To investigate the hypothesis that body mass index is associated with lung cancer independent of cigarette smoking, the authors analyzed data from a prospective cohort study of 41,837 Iowa women aged 55-69 years at baseline in 1986. In addition, they examined whether central adiposity (high waist/hip ratio) was associated with lung cancer incidence. Through 1992 (6 years of follow-up), 233 cases of lung cancer were identified through the State Health Registry of Iowa. The body mass index at several ages was calculated from self-reports of height at baseline and weights at ages 18, 30, 40, and 50 years and at baseline. Current and former smokers generally had lower mean body mass indices than did nonsmokers at all ages except 18 years. Cases generally had lower body mass indices than did noncases at all ages except 18 and 30 years but, among current smokers, cases had higher mean body mass indices than did noncases at all ages except baseline, although the differences were not statistically significant. An apparent positive association of a high waist/hip ratio with lung cancer in the total cohort was found to be primarily accounted for by a higher waist/hip ratio in current and former smokers. When stratified by smoking status and adjusted for other risk factors, including age and pack-years of smoking, the body mass index at baseline, body mass index at age 50 years, and waist/hip ratio were not associated with lung cancer. The results of multivariate analyses suggest that the inverse association of body mass index with lung cancer can be explained by smoking status and that the positive association of waist/hip ratio with lung cancer can be explained by pack-years of smoking.