Body composition appears to be an important determinant of bone mineral density (BMD). BMD at the femoral neck, lumbar spine, and whole-body and the whole-body soft-tissue composition were measured cross-sectionally in 334 healthy premenopausal and early perimenopausal women, aged 44-50 years, using a Hologic QDR densitometer. Correlations between lean mass and BMD at the hip, spine, and whole-body were greater (r = 0.40, r = 0.44, and r = 0.45, respectively, p < 0.0001) than those for fat mass (r = 0.19, r = 0.16, and r = 0.16, respectively, p < 0.01). There was a significant linear trend in femoral BMD from the lowest to highest category of lean mass (0.75 +/- 0.10 g/cm2, 0.80 +/- 0.10 g/cm2, and 0.86 +/- 0.09 g/cm2, p < 0.0001). Similar trends were demonstrated for spinal and whole-body density. For categories of fat mass, there was a significant linear trend at the hip (0.78 +/- 0.10 g/cm2, 0.79 +/- 0.10 g/cm2, and 0.83 +/- 0.10 g/cm2, p = 0.0106), but not at the spine or whole body. There was a 5.00% (3.62, 6.38; 95% confidence limits) difference in hip BMD per unit (standard deviation) of lean mass, while only a 0.73% (-0.66, 2.11) difference in hiP BMD per unit (SD) of fat mass. Differences in BMD were examined by categories of lean and fat mass (low, medium, high) for a total of nine possible combinations of lean and fat measures. BMD at the hip, spine, and whole-body were significantly higher in those with high lean mass than in those with low lean mass, irrespective of fat mass. Women with high lean/low fat had similar hip, spinal, and whole-body BMD as those with high lean/high fat, despite their significantly lower body weight (62.5 +/- 3.3 kg vs 85.7 +/- 5.4 kg, respectively, p < 0.0001). In premenopausal and early perimenopausal women, body weight alone may not be associated with increased bone mass unless a significant proportion of that weight is comprised of lean mass. The stronger association between lean mass and BMD than that for fat mass may be attributed to differences in determinants of lean mass, such as exercise, lifestyle factors, estrogen levels, or a combination of these factors.