Estimating body composition is important in assessing the progress of performance potential or physical condition in athletes in training. However, convenient and accurate methods for athletes with highly trained physical characteristics have not yet been established. Meanwhile, somatotyping has been used for the assessment of athletes' physiques. To investigate the utility of somatotyping for estimating body composition by clarifying relations of somatotype components and body composition indexes in Japanese college athletes. Measurements were made in thirty-one male college track and field athletes (18-22 yr) two times each, at 13-month intervals. We measured weight, height, breadth and girth measurements, skinfold thicknesses (SF), and bioelectrical impedance (BI). Three somatotype components (i.e. endomorphy, mesomorphy and ectomorphy) were determined depending on the Heath-Carter anthropometric method. Body mass index (BMI; kg·m- 2), and fat mass index (FMI; kg·m−2) and fat-free mass index (FFMI; kg·m−2) (i.e. fat mass and fat-free mass standardized by height), calculated from both SF and BI (FMI-SF, FMI-BI, FFMI-SF and FFMI-BI, respectively), were used as body composition indexes. The relation between somatotype and body composition was analyzed for all two measurements. The relations of each somatotype component and body composition index are listed in Table 1. Mesomorphy and ectomorphy were significantly related to BMI. FMI and FFMI by SF were more accurate than by BI in relation with somatotype components. High correlations were obtained in the endomorphy vs. FMI, mesomorphy vs. FFMI, and ectomorphy vs. FFMI relations. No matter what estimation method for body composition was used, each somatotype component was significantly related to body composition. These findings in the present study suggest that anthropometric somatotype rating would be useful for estimating body composition in athletic training.