Bone mineral density in adult life, which is an important determinant of fracture risk, is determined by peak adult bone density, achieved in early adulthood and subsequent rates of change during adult life. Cross-sectional twin and family studies indicate that the majority of population variation in bone density may be explained by genetic factors. Although there is evidence for a genetic effect on peak bone mass, it is unknown whether there is a genetic effect on rates of changes in bone density with age. Changes in lumbar spine and femoral neck bone density determined by dual-photon absorptiometry (Lunar DP3) were examined in a cohort of monozygotic (MZ, n = 21, 3 male and 18 female pairs, median age, range, 46; 24-75 years) and dizygotic twins (DZ, n = 19; 43, 25-65 years). The median follow-up was 3 years (range 1.1-5.5 years), with each subject having at least two and up to four bone density assessments. In these twins, genetic factors determine variation in rates of change (% change/year) in lumbar spine bone density, rMZ = 0.93 and rDZ = 0.51, p < 0.02 (one tailed), and Ward's triangle, rMZ = 0.60, rDZ = 0.11, p < 0.05 (one tailed). Model-fitting analysis was also consistent with a genetic effect on rates of change in bone density at the trochanteric site, although such an effect was not shown at the femoral neck. These data demonstrate, for the first time, the possible existence of genetic determinants of rates of change in bone mineral density in adults.