To investigate the effects of isoflavone supplementation on both bone mineral density and hormone variation in premenopausal women who had decreased bone mass, the 24 subjects were divided into two groups: one was the underweight group, consisting of 13 subjects, and the other was the normal weight group consisting of 11 subjects. For each group, we investigated the effects of isoflavone supplementation of 90 mg/day on both bone mineral density and hormone variation during 3 menstrual cycles. Anthropometric measurements, dietary recall, and analyses of blood and urine were assessed from baseline to post-treatment. The results were as follows: The average age of the underweight group was 21.8 years old and that of the normal weight group was 23.2 years old. The comparative results for the two groups at baseline were as follows: Onset of menarche, menstrual cycle, and menstrual length were not significantly different between the groups. Serum protein, total, HDL-, LDL-cholesterol, triglyceride, Ca, P, Mg, Cu, and Zn level were not significantly different between the groups. Serum estradiol, SHBG, LH, and FSH level were also not significantly different between the groups. Lumbar spine BMD by T scores of the underweight group was significantly lower than that of the normal weight group. Serum osteocalcin, urinary DPD, and urinary pH were not significantly different between the groups. The comparative results for the two groups at post-treatment were as follows: From baseline to post-treatment, the intake of energy, nutrients and isoflavone in food did not significantly change in either group. Serum protein, total cholesterol, HDL-, LDL-cholesterol, and triglyceride levels did not significantly change in either group. Serum Ca, Cu, and Zn levels were significantly lower in both groups and serum Mg level significantly decreased only in the underweight group. Serum estradiol levels were significantly lower in both groups, but serum SHBG, LH, and FSH levels did not significantly change in either group. Lumbar spine BMD by T score of the underweight group significantly increased to 15%, but that of the normal weight group did not significantly change. Serum osteocalcin of the underweight group significantly increased to 28%, while that of the normal weight group significantly increased to 40%. Urinary DPD of the normal weight group significantly increased to 12%. The results show that the BMD of the underweight group was lower than that of the normal weight group. Therefore, the underweight group had a disadvantage in obtaining maximum bone mineral density. The results also show that isoflavone supplementation during 3 menstrual cycles was effective in increasing the bone mineral density of the lumbar spine and affected bone metabolism markers in premenopausal underweight women. Therefore, it can be concluded that sufficient intake of isoflavone could be helpful in preventing decreases in bone mass among premenopausal women, especially underweight women.