The racial groups were American Blacks (n=320), American Whites (n=672), Hispanics (n=41), Native Americans (n=1,659), and Polynesians (n=179). The environmental variables were subsistence strategy, physical terrain, and geographical region. Four femural measurements were obtained for mature individuals in these groups: femur maximum length, vertical head diameter, subtrochanteric anteroposterior, and mediolateral diameter. The samples were measured by multiple observers, but no test of interobserver measurement error could be conducted. The study found significant within-group variation in proximal femur shape. Among Native Americans, both geographic location and subsistence strategy (agriculturalist, plains horticulturalist, or hunter-gatherer) had a significant effect on proximal femur shape; nevertheless, the study generally verified the belief that the proximal femur can be reliably used to distinguish Native Americans from American Blacks and American Whites, but precision may be reduced in some geographical regions. Measurements of Hispanics did not differ significantly from American Blacks and American Whites. 9 tables, 1 figure, and 61 references