In this cross-sectional study, bone mass and muscle strength of the thigh were investigated in 51 female soccer players, age 16.3 +/- 0. 3 years, who had been playing soccer for 8.1 +/- 2.1 years and were at the time of the study in soccer training for 5.0 +/- 1.7 hours/week. They were compared with 41 nonactive females, age 16.2 +/- 1.3 years. The groups were matched according to age, weight, and height. Areal bone mineral density (BMD) was measured of the total body, head, lumbar spine, femoral neck, Ward's triangle, and the greater trochanter using dual energy X-ray absorptiometry (DXA). Isokinetic muscle strength of the quadriceps and hamstrings muscles was measured using an isokinetic dynamometer. Compared with the nonactives, the soccer players had significantly higher BMD of the total body (2.7%), lumbar spine (6.1%), the dominant and nondominant hip (all sites). The largest differences were found in the greater trochanter on both sides (dominant, 16.5%, nondominant, 14.8%). The soccer players had significantly higher concentric and eccentric peak torque of the thigh muscles. In the soccer group, there was only a positive association between thigh muscle strength and BMD of the adjacent hip, and in the nonactive group there were several positive associations between muscle strength and BMD. However, when adjusting for the variation in weight and height all these associations became nonsignificant. Using multiple linear regression, the type of activity (soccer player, nonactive) independently predicted BMD of all dominant hip sites (beta = 0.32-0.48, P < 0.01). No other variable was found to independently predict BMD of any site. In the younger subjects (</=16 years) only BMD of the greater trochanter was significantly higher in the soccer players. In the older subjects (>16 years) the soccer players had significantly higher BMD in all measured sites except for the nonweight-bearing head. The differences in muscle strength between soccer players and nonactives were already seen in the young age group. In conclusion, girls who train and play soccer in adolescence have a higher bone mass in the hip and lumbar spine, and a higher muscle strength of the thigh compared with nonactive controls, indicating a site-specific skeletal response of weight-bearing and impact-loadng acting on the skeleton. The differences in bone mass were already apparent in early adolescence, but became more pronounced in late adolescence, probably explained by a longer exposure to soccer training with time. Our results also indicate that muscle strength in itself might not be of decisive importance for bone mass in the hip of adolescent females.