Potential determinants of bone mineral density (BMD) were studied cross-sectionally in 115 healthy, sexually mature Caucasian women aged 18 years. Bone mineral density (Hologic QDR1000W) of the lumbar spine, proximal femur (five sites), and distal tibia and fibula; fasting blood and urine calcium biochemistry; serum sex hormone levels (follicular phase); nutrient intakes; aerobic fitness; trunk muscle strength; and habitual activity levels were measured. The effects of heredity were considered by measuring the BMD of 107 of the subjects' mothers. Simple and stepwise regression analysis were used to identify significant determinants of BMD at each of the regions studied. The analysis indicated that significant bivariate correlations exist between BMD at all sites and body weight (r = 0.23-0.47, p < or = 0.01), lean body weight (r = 0.34-0.46), trunk strength (r = 0.27-0.47), physical activity score (r = 0.20-0.25), and aerobic fitness (r = 0.29-0.45). Dietary calcium intake correlated significantly with BMD at the trochanter site only (r = 0.19), and none of the biochemical or hormonal indices measured correlated consistently with BMD at any site. Significant correlations between the BMD of mothers and daughters ranged from r = 0.43 at lumbar spine to r = 0.34 at the intertrochanteric site. Paired t-tests showed the daughters had significantly (p < 0.03) lower BMD than their mothers at the lumbar spine (98 +/- 12% [mean +/- SD]) and significantly higher (p < 0.002) BMD at the femoral neck, trochanter, and total hip sites (110 +/- 16%, 108 +/- 17%, 103 +/- 14%, respectively). When stepwise regression analysis included weight-corrected strength of the trunk flexor muscles (Corr Flex), weight-corrected aerobic fitness (Corr VO2max), physical activity score, and body weight, body weight was the only significant determinant of BMD at all sites. Corr Flex made significant contributions at all sites except the femoral neck, while Corr VO2max made additional contribution at the femoral neck, trochanter, total hip, and shaft of femur sites. These variables accounted for 13-27% of the variance in BMD. The addition of mother's BMD to these independent variables, in stepwise regression analysis, improved the prediction to 18-31% of the variance.