The aim of this study was to investigate the effect of training and detraining on bone mineral density of both weight-bearing and non-weight-bearing bone in a cohort of young males who participated in ice hockey training. Forty-three healthy adolescent ice hockey players (16.7+/-0.6 years) training for a mean of 9.7+/-2.4 h/week and 25 control subjects (16.8+/-0.3 years) training for 2.1+/-2.7 h/week, were included in this longitudinal study. Bone mineral density (BMD, g/cm2) of the arms, the dominant and non-dominant humerus, dominant and non-dominant femur, and the right femoral neck, total hip, and bone area of the femur, humerus and hip were measured at baseline and again after 30 and 70 months using dual-energy X-ray absorptiometry. From baseline to the first follow-up, athletes gained significantly more BMD in the femoral neck (0.07 versus 0.03 g/cm2) and arms (0.09 versus 0.06 g/cm2) compared with the controls (P = 0.04 for both). Between the first and the second follow-up, 21 ice hockey players stopped their active sports career. These men lost significantly more BMD at the femoral neck (-0.02 versus -0.10 g/cm2, P < 0.001), total hip (-0.05 versus -0.09, P = 0.04), dominant (0.02 versus -0.03 g/cm2, P = 0.009) and non-dominant humerus (0.03 versus -0.01 g/cm2, P = 0.03) than the still active ice hockey players (n = 22). At the second follow-up examination, at 22 years of age, the former ice hockey players still had significantly higher BMD at the non-dominant humerus than the controls (P < 0.01). During the total study period, the still active athletes (n = 22) gained significantly more BMD compared with the controls at the femoral neck (0.09 g/cm2; P = 0.008), total hip (0.05 g/cm2, P = 0.04) and arms (0.07 g/cm2; P = 0.01). No differences were seen in bone areas when comparing the different groups. In conclusion, training associated with ice hockey is related to continuous accumulation of BMD after puberty in males. Reduced activity is followed by BMD loss within 3 years of cessation of sports career at predominantly weight-bearing sites. The effects are confined to bone density and not bone size.