Few studies examine the relationships between anthropometry and the body composition measures they approximate, or whether they differ by sex, and no studies have examined these relationships in South Asians living in the US.We conducted a cross-sectional study of 871 participants in the Mediators of Atherosclerosis in South Asians Living in America (MASALA) Study who had BMI < 40 kg/m2 and underwent abdominal CT scans for measurement of visceral and subcutaneous fat. Linear regression was used to model the associations between anthropometric measures and naturally log-transformed body composition measures.All measures of anthropometry, except height, were significantly associated with visceral fat and had a significant non-linear component (p < .05). The only associations for visceral fat that exhibited significant heterogeneity by sex were waist circumference (% difference in visceral fat slope: women 1.92, men 2.74, p = .007 for interaction) and waist-to-hip ratio (women 25.9, men 717.4, p < .001). Except for height, all measures of anthropometry were significantly associated with subcutaneous fat, had a significant quadratic component, and significant heterogeneity by sex (weight (kg): 2.74 for women, 4.08 for men; BMI (kg/m2): 10.3, 14.0; waist circumference (cm): 1.51, 3.36; hip circumference (cm): 2.53, 4.50) with p < .001 for each.In MASALA participants, the relationships of anthropometric measures with visceral and subcutaneous fat appear similar to other race/ethnic groups, but with weaker non-linearity and heterogeneity by sex. Given these results, researchers should consider separate models by sex for US South Asians when approximating subcutaneous fat or when using waist circumference to approximate visceral fat.