Weight-bearing exercise during growth enhances peak bone mass. However, the window of opportunity for optimizing positive effects of exercise on peak bone mass remains to be fully defined. Ballet dancing provides a model of mechanical loading patterns required to site-specifically modulate bone.We assessed the effects of ballet dancing on bone mineral accrual in female non-elite dancers and normally active controls for 3 years across puberty. We recruited 82 ballet dancers and 61 controls age 8-11 years at baseline. Participants were measured over 3 consecutive years; however, the overlap in ages allowed analysis of the groups across 8-14 years of age. We annually assessed bone mineral content (BMC) at the total body (TB), including upper and lower limb regions, and biannually assessed BMC at the proximal femur and lumbar spine (LS) using dual x-ray absorptiometry (DXA). We derived TB lean mass and fat mass from DXA TB scans. Anthropometry, exercise levels, and calcium intake were also measured biannually. Maturational age was determined by age at peak height velocity (PHV). A multilevel regression model was used to determine the independent effects of body size, body composition, maturation, exercise levels, and calcium intake at each measurement occasion.When adjusted for growth and maturation, dancers had significantly greater BMC at the TB, lower limbs, femoral neck (FN), and LS than controls. Excepting the FN region, these differences became apparent at 1 year post-PHV, or the peripubertal years, and by 2 years post-PHV the differences represented a cumulative advantage in dancers of 0.6-1.3% (p<0.05) greater BMC than controls. At the FN, dancers had 4% (p<0.05) greater BMC than controls in prepuberty and maintained this advantage throughout the pubertal years.Results from this novel population provide evidence for modest site-specific and maturity-specific effects of mechanical loading on bone.