The question of whether and to what extent the sex steroid deficiency in elderly men contributes to the pathogenesis of bone loss has not been fully explored. The aim of the present study was to assess the association of serum bioavailable (Bio) estradiol (E(2)) with the evolution of bone mineral density (BMD) in 214 community-dwelling men aged 71-86 yr as well as the possible modulation of estrogen effects by a tetranucleotide (TTTA)(n)-repeat polymorphism of the CYP19 gene, which encodes the aromatase enzyme that converts androgens into estrogens. BMD was measured at yearly intervals over a period of 4 yr using dual x-ray absorptiometry. Fasting blood was analyzed at baseline for testosterone (T), E(2), and SHBG; the respective bioavailable fractions, BioT and BioE(2), were calculated. Serum BioE(2) was associated with baseline BMD at different assessed skeletal sites, with correlation coefficients ranging between 0.23 and 0.37 (P < 0.001). Estimated annual percentage change of BMD (%BMD) was -0.39% [95% confidence index (CI), -0.56, -0.22] at the total hip, -0.04% (95% CI, -0.29, 0.21) at the femoral neck, and -0.37% (95% CI, -0.45, -0.29) at the total distal forearm. Higher circulating BioE(2) levels were associated with less bone loss at the forearm and the hip (P < 0.05). The CYP19 gene (TTTA)(n)-repeat length (determined by fragment analysis) was not associated with baseline BMD in the total group of elderly men. However, a significant association was observed between the CYP19 genotype and BMD change at the distal forearm; the highest bone loss was observed in subjects homozygotic for the shortest observed allele length of (TTTA)(7)-repeats (P < 0.02). The CYP19 (TTTA)(n)-repeat length was not associated with either baseline BioE(2) or the BioT/BioE(2) ratio. In multiple linear regression models, the CYP19 genotype and serum BioE(2) were determinants of %BMD change at the forearm (P < 0.05). No significant contribution of BioT to %BMD change was evident. As to fracture risk, the allele containing the shortest (TTTA)(n)-repeat length was more represented not only in elderly men with a positive personal fracture history (Pearson's chi(2) test = 4.03; df = 1; P = 0.05) but also in study subjects with a positive fracture history in their first-degree relatives (Pearson's chi(2) test = 6.48; df = 1; P = 0.01). In conclusion, the results of this prospective observational study support the view that BioE(2) is a determinant of bone density changes in elderly men and, furthermore, provide an indication that the aromatase enzyme may exert a direct modulatory action on bone metabolism at the tissue level in elderly men.