Lean mass and areal bone mineral density at the lumbar spine, femoral neck, and total forearm were measured in 215 volunteer female twin pairs (122 monozygotic, 93 dizygotic), aged 10–26 years, using dual energy X-ray absorptiometry. The study was conducted in Melbourne from 1990 to 1994. Under the classic twin model, there was evidence for a genetic component of variation in bone mineral density, adjusted for age or for age and lean mass, at all sites. Adjusting for lean mass almost halved the genetic variances in the adolescent years of peak growth, during which genetic variances peaked. Genetic variances were reduced in the late teenage years and increased in early adulthood. The latter may reflect gene-environment interactions or covariation. Importantly, there was evidence for environmental effects shared by twins on lumbar spine and femoral neck bone mineral density, even when adjusted for lean mass as well as age. These were greatest during the late teenage years, abated over the years when pairs started to live apart, and appear to be independent of lean mass during adolescence but not in early adulthood. In summary, the genetic and environmental etiology of bone mineral density is more complex than previously thought.