To assess whether risk of breast cancer in young women is associated with differences in luteal-phase hormone production and to attempt to explain differences in risk of breast cancer of young Shanghai Chinese and Los Angeles white women, two concurrent case-control studies of serum hormone concentrations were conducted. Both studies were carefully controlled for the possible confounding effects of age, weight, height, pregnancy history, and day of the menstrual cycle, by individually matching cases and controls on these factors. Case eligibility was limited to women with localized breast cancer. Sixteen of 39 Shanghai breast-cancer cases were sampled prior to the histologic diagnosis of their disease. The remaining 23 Shanghai cases and all 42 Los Angeles cases were diagnosed, and treated by surgery only, at least six months prior to hormonal evaluation. All subjects were sampled on day 22 of the menstrual cycle. Overall, cases had 13.5% higher serum estradiol concentrations (p = 0.038) with a case-to-control excess of 16.6% in Shanghai subjects (p = 0.089) and 10.8% in Los Angeles subjects (p = 0.23). There were no appreciable differences in amounts of sex-hormone binding globulin between cases and controls. Cases had lower progesterone levels than controls, but the situation was reversed when the analysis was restricted to subjects with evidence of ovulation. Los Angeles controls had 20.6% greater estradiol concentrations than Shanghai controls (p = 0.036); adjustment for body weight accounted for only 25.7% of this difference.(ABSTRACT TRUNCATED AT 250 WORDS)