Bone mineral density (BMD) was studied in young exercising amenorrheic girls to determine if density was compromised and the change related to injury. Ninety-eight volunteers (professional ballet dancers and controls) were studied in a cross-sectional study. Dancers and controls were further subdivided into normally cycling and amenorrheic subjects. Amenorrhea significantly lowered bone density of the spine (P less than 0.0001), wrist (P less than 0.03), and metatarsal (P less than 0.01); effects on the wrist were eliminated by controlling for age while controlling for weight eliminated all effects of amenorrhea at three sites. BMD of the metatarsal, a weight-bearing bone, showed an interaction between amenorrhea and dancing (exercising) P less than 0.035); surprisingly, dancing was associated with a further lowering of bone density. This interaction was eliminated when controlling for age, but not when controlling for weight. With multiple comparisons of the groups, spine, wrist, and metatarsal bone density was significantly lower in amenorrheic dancers when compared to normal dancers (P less than 0.05), even when controlling for age and weight in the metatarsal (P less than 0.05), and age in the spine (P less than 0.05). Estradiol levels correlated with bone density of both the wrist and the spine (r = 0.25, r = 0.23, P less than 0.02). Metatarsal density correlated with estradiol levels only in the dancers (r = 0.34, P less than 0.02). The only variable found to correlate with the occurrence of stress fractures was age of menarche. This was also the only variable of 9 (BMD of the wrist, spine or foot, calories ingested and expended, amount of calcium ingested, involvement in high energy activity, age of menarche or presence of amenorrhea) to predict stress fractures. Thus, BMD is significantly affected by the presence of amenorrhea but the effects are generally weight dependent. The compensatory increase in bone density generally seen in stressed bones, such as the metatarsal in ballet dancers, is deficient in amenorrheic premenopausal women even when controlling for weight but this effect may be age and estrogen dependent. Bone mass may not accumulate in the same manner in adolescents as in the mature women, thus putting them at risk for injury.