Somatotypes of a cross-sectional sample of 544 rural adolescents ranging in age from 11 to 17 years are described. The sample included 269 Rajput (141 girls and 128 boys) and 275 Scheduled Caste (135 girls and 140 boys) subjects. Each subject was somatotyped using the Heath-Carter anthropometric somatotype protocol (Carter & Heath 1990). In all, ten anthropometric measurements namely height, weight, bicondylar diameters of humerus and femur, flexed mid-upper-arm and calf circumferences, and triceps, subscapular, supraspinale and calf skinfolds were taken. The mean somatotypes of the Rajput boys and girls were 1.62- 3.30-3.85 (mesomorphic-ectomorph) and 2.42-2.90-3.99 (balanced ectomorph), respectively. The mean somatotypes of the Scheduled Caste subjects were 1.51-3.02-3.74 (mesomorphic-ectomorph) for boys and 2.38-2.64-3.70 (balanced ectomorph) for girls. A one-way ANOVA revealed that females of both the caste groups were significantly (p < or = 0.05) more endomorphic than the males. The sex differences in other two components were not significant (p +/- 0.05). Caste differences, as revealed by a one-way ANOVA analysis, were not significant (p +/- 0.05) in both sexes. With the exception of the Rajput girls, the differences in whole somatotypes between those in an early phase of adolescence and those in an advanced phase of adolescence were not significant (p = 0.05). The results indicate that populations exposed to same environmental situations for a long period of time tend to show similarity in physique. A one-way MANOVA analysis, which used Wilk's Lambda as test statistics, revealed that from 11-17 years there was no significant change (p < or = 0.05) in component dominance of mean somatotypes in the boys and girls of the present sample. Among males of a majority of the Indian populations, ectomorphy dominates over endomorphy and mesomorphy from 11 to 17 years.