OBJECTIVE: Test the hypothesis that soy isoflavone supplementation preserves bone mineral density (BMD) in men and women. METHODS: We conducted a controlled, parallel-arm, double-blinded trial with 145 participants, 50-80 years, with random assignment to soy beverage daily for 12 months. Active treatment (+ISO) received soy protein containing 83 mg isoflavones (45.6 mg genistein, 31.7 mg daidzein), aglycone units; the comparison group (-ISO) received soy protein containing 3mg isoflavones. We measured BMD using dual-energy X-ray absorptiometry at the total hip and posterior-anterior spine (L1-L4) at baseline in 22 women and 123 men, and at 12 months in 13 women and 98 men. We used linear mixed models to test for an isoflavone effect on percentage BMD change from baseline in spine and hip. RESULTS: Among all participants, mean percent change in spine BMD (+/-S.E.) was 0.16+/-0.44 in -ISO (P=0.10) at 12 months. Treatment effects on spine BMD were significantly greater in women than men (P=0.01). At 12 months, in women, mean percent change was 0.58+/-0.70 in +ISO and -1.84+/-0.86 in -ISO (P=0.05); among men it was 1.32+/-0.53 in +ISO and 0.31+/-0.48 in -ISO (P=0.16). By comparison, percent change in hip BMD was similar in the treatment groups, and was not different between men and women. Mean percent change in hip BMD from baseline to 12 months was 0.54+/-0.38 in +ISO and -0.13+/-0.36 in -ISO (P=0.20) among all participants. CONCLUSIONS: Soy protein containing isoflavones showed a modest benefit in preserving spine, but not hip BMD in older women.