It has been proposed that the high intake of soy foods among Asians may partly explain their lower rates of breast cancer, perhaps by lowering endogenous oestrogen levels, although this has been inadequately studied. Twenty healthy cycling premenopausal women (ten Asians and ten non-Asians) participated in a 7-month soy intervention study which was designed to investigate the effect of supplementation on ovarian function. Asian soy foods (tofu, soymilk, green soybean peas) in the amount of approximately 32 mg of isoflavones per day were added to the women's diets for three menstrual cycles. The women's baseline (two cycles) serum hormone levels were compared to levels during soy intervention (three cycles) and levels after intervention (two cycles). During the entire study period, subjects provided almost daily overnight urine samples and blood specimens during specified days of their menstrual cycles. The day of urinary luteinizing hormone (LH) peak was used as a marker for the day of ovulation. Knowledge of day of ovulation allowed comparison of hormone measurements at baseline to those obtained during intervention and recovery cycles with standardization of day of cycle. Soy intervention was associated with a statistically significant reduction in serum luteal oestradiol level (–9.3%,
P
< 0.05), but there were no significant changes in follicular phase oestradiol, follicular or luteal phase progesterone, sex hormone-binding globulin or menstrual cycle length. This significant reduction in luteal phase oestradiol was, however, observed only among Asian (–17.4%) but not among non-Asian (–1.2%) participants; urinary excretion of isoflavones was higher among Asians than non-Asians (29.2 vs 17.1 μmol day
−1
,
P
= 0.16) during the intervention period. Thus, supplementation using traditional soy foods reduced serum oestradiol levels among Asian participants in this study. Differences in the type of soy products (i.e. traditional soy foods versus soy protein products), amount of isoflavones, and race/ethnicity of participants may have contributed to the divergent results. Larger soy intervention studies designed specifically to include participants of different race/ethnicities and using both traditional soy foods and soy protein products providing comparable doses of isoflavones are needed to definitively determine the effect of soy on ovarian function. © 2000 Cancer Research Campaign