A longitudinal design was used to study the development of sprint performance in eight first-class male sprinters from the age of 12 to 15 years. Measurements were performed annually every November for sprint speed, step frequency, step length, sprint motion, and isokinetic peak torques during knee extension and flexion. Furthermore, attempts were made to analyze support time and non-support time, angular kinematics, and isokinetic peak torques at 60, 180, 300 deg/s. The results obtained were as follows : 1. At the ages of 12 and 13 years, mean values of body height and weight were larger for the subjects than for normal Japanese boys. However, at the last measurement at the age of 15 years, no differences were found between them. The skeletal age of the subjects was about 3 years more advanced than chronological age at 12 years. These data therefore, confirmed that the subjects showed an earlier growth spurt. 2. From 12 to 14 years of age, sprint speed increased significantly from 8.79 m/s to 9.61 m/s. From 12 to 13 years of age, step length increased significantly from 1.95 m to 2.08 m. However, no significant increases in step frequency were observed with age. From 12 to 13 years of age, the improvement of sprint speed resulted from the increase in step length rather than step frequency. 3. Significant increases were seen in maximal thigh lift velocity (ωT) from 12 to 14 years of age and in maximal leg swing velocity (ωL) from 12 to 13 years of age. However, no correlations were found between the increases in sprint speed and those of all sprint motions such as maximal leg swing velocity from 12 to 15 years of age. 4. From 12 to 13 years of age, isokinetic peak torques during knee flexion (angular velocity; 60, 180, 300 deg/s) showed significant increases. From 12 to 15 years of age, increases in sprint speed were significantly correlated with those of isokinetic peak torque during knee flexion at angular velocities of 180 and 300 deg/s. For all ages, significant partial correlations were found between sprint speed and isokinetic peak torque during knee flexion at angular velocities of 180 and 300 deg/s. 5. In spite of the fact that one subject (IK) showed better sprint performance than another subject (NI) at the age of 12 years, the latter subjects was able to perform better than the former at the age of 15 years. This change was due to the larger increase in body height and weight as well as the greater development of isokinetic peak torque during knee flexion in NI than in IK. 6. The relationship between the increase in sprint speed and skeletal maturity indicated that the increase in individuals who were skeletally immature after the age of 12 years were greater than those in individuals who matured early.