The likelihood of low trauma fracture in the elderly is highly predictable by peak bone mass (PBM) at age approximately 25-50 yr. We estimated the magnitude of genetic determination of the variation and covariation of PBM of the spine and hip (adjusted by age, gender, and ethnicity) in 47 independent healthy full-sib pairs and 27 healthy mother-offspring pairs. For the spine and hip, the narrow-sense heritabilities (h(2)) (mean +/- SE) were 0.76 +/- 0.34 and 0.84 +/- 0.36, respectively, when estimated from full sibs, and 0.86 +/- 0.38 and 0.84 +/- 0.39, respectively, when estimated from parent-offspring. Some genetic loci underlying PBM variation at the hip and spine are the same or closely linked, as is reflected by the high genetic correlation of 0.95 +/- 0.05 between them when estimated from full sibs, and 0.57 +/- 0.27 when estimated from parent-offspring, respectively. Generally, common familial environmental effects shared by relatives may bias these estimates. However, these effects may be small, since our results reported herein and those in other earlier studies indicate that common familial environmental effects are probably negligible in causing similarity of bone mass among family members. The correlation of bone mass among randomly sampled couples living in the same household is small and nonsignificant as measured either by densitometry at the radius and ulna or by quantitative ultrasound at the patella. The problem of shared environmental effects notwithstanding, we conclude that much of the PBM variation and covariation at the hip and spine is determined genetically.