A lower BMR of Indians, when compared with Westerners matched for age, sex and either surface area or body weight, has often been reported in the literature and has been interpreted to reflect an ethnic influence on BMR. To determine the contribution of body composition to these observed differences in BMR, we analysed the data on ninety-six Indians and eighty-one Caucasian Australians of both sexes, aged 18-30 years, studied in Bangalore, India and Melbourne, Australia. Absolute BMR and BMR adjusted for body weight were significantly lower in Indians when compared with Australians of the corresponding sex. However, BMR adjusted for fat-free mass (FFM) in men, and BMR adjusted for FFM and fat mass (FM) in women, were not significantly different between the two groups. Stepwise regression of FFM, FM, sex (0 = women; 1 = men) and ethnicity (0 = Indian; 1 = Australian) on BMR, resulted in the following relationship for the combined data on all subjects: BMR = 88.7 x FFM (kg) + 1713 (n 177; r 0.92; r2 0.85; SEE 425 kJ). The Indian equations of Hayter & Henry (1994), based on body weight, resulted in a significant bias (measured-predicted BMR) of 318 (SE 54) kJ/d in Indian men and -409 (SE 70) kJ/d in Indian women. The equation of Cunningham (1991), based on FFM, accurately predicted the BMR of Indian men, Indian women and Australian men. The small but significant bias of 185 (SE 61) kJ/d in Australian women, may be explained by the significant contribution of FM to BMR in this group. The present study does not provide any evidence for an ethnic influence on basal metabolism. The results strongly support the use of FFM, rather than body weight, for the prediction of BMR in population groups of varying body size and composition. This would allow an accurate estimation of BMR and hence energy requirements in population groups worldwide.