Data from a case-control study on the upper digestive and respiratory tract cancers were analyzed to investigate the relationship with anthropometric measures. The data set included 538 oropharyngeal, 410 esophageal, and 388 laryngeal cancer cases and 2,102 controls in hospital for acute nonneoplastic diseases, unrelated to tobacco or alcohol use, recruited in the same catchment areas as the cases. Lower body weight appeared to be an indicator of oropharyngeal and esophageal cancer and, more moderately, of laryngeal cancer. The multivariate odds ratio (OR) for oropharyngeal cancer was 5.0 for subjects in the lowest compared with the highest quartile of weight. The corresponding ORs were 6.2 for esophageal and 2.2 for laryngeal cancer. When quartiles of body mass index (BMI) were considered, the ORs in the lowest quartile were 4.0 for oropharyngeal, 6.3 for esophageal, and 2.4 for laryngeal cancer. Subjects in the lowest quartile of height had ORs of 2.0 for oropharyngeal, 1.6 for esophageal, and 1.4 for laryngeal cancer. When all the upper digestive and respiratory tract neoplasms were considered together, the OR for subjects with lower BMI who were also current smokers compared with never-smokers with higher BMI was 11.4. The OR was 5.0 for those consuming six or more drinks/day and with low BMI. Although the nature of the association needs to be clarified, these data suggest that leanness may be involved in the process of upper digestive and respiratory tract carcinogenesis.