In a cross-sectional epidemiological study including 3290 men aged 53-75 y, mean ¼ 63, we tested the hypothesis that the Lewis phenotype Le(aÀbÀ) is a genetic marker of obesity. All men were Lewis typed and measurements were made of height, weight, neck circumference, and hip fat fold. Obesity was defined as a body mass index Z30 kg/m 2 . Totally 291 men (8.8%) were obese; 9.6% of the men had the Le(aÀbÀ) phenotype. Le(aÀbÀ) men had a higher prevalence of obesity than others, 15.6 vs 8.1%, odds ratio (95% confidence interval): 2.1(1.5-2.9), Po0.001. The etiological fraction, that is, the excess prevalence of obesity in the study population due to the Le(aÀbÀ) phenotype, was approximately 10%. The frequency of the Le(aÀbÀ) phenotype may vary substantially in different populations. Identification of this new genetic marker of obesity may, for example, contribute to the explanation of individual and ethnic differences in the prevalence of obesity.