OBJECTIVE - Physical activity and insulin sensitivity are related in epidemiological studies, but the consistency of this finding among populations that greatly differ in body size is uncertain. The present multiethnic epidemiological study examined whether physical activity was related to insulin concentrations in two populations at high risk for diabetes that greatly differ by location, ethnic group, and BMI. RESEARCH DESIGN AND METHODS - The study populations consisted of 2,321 nondiabetic Pima Indian men and women aged 15-59 years from Arizona and 2,716 nondiabetic men and women aged 35-54 years from Mauritius. Insulin sensitivity was estimated by mean insulin concentration (average of the fasting and postload insulin), and total (i.e., leisure and occupational) physical activity was assessed by questionnaire. RESULTS -Pima men and women who were more active had significantly (P < 0.05) lower mean insulin concentrations than those less active (BMI and age-adjusted means were 179 vs. 200 and 237 vs. 268 pmol/l). Similar findings were noted in Mauritian men and women (94 vs. 122 and 127 vs. 148 pmol/l). In both populations, activity remained significantly associated with mean insulin concentration controlled for age, BMI, waist-to-thigh or waist-to-hip ratio, and mean glucose concentrations. CONCLUSIONS - Physical activity was negatively associated with insulin concentrations both in the Pima Indians, who tend to be overweight, and in Mauritians, who are leaner. These findings suggest a beneficial role of activity on insulin sensitivity that is separate from any influence of activity on body composition.