To assess the relation between urinary endogenous sex steroid levels and the risk of postmenopausal breast cancer, a nested case-cohort study was conducted within a large cohort (the DOM cohort) in the Netherlands (n=9,349). Until the end of follow-up (1 January 1996), 397 postmenopausal breast cancer cases were identified and a subcohort of 424 women was then taken from all eligible women. Women using hormones were excluded, leaving 364 breast cancer cases and 382 women in the subcohort for the analyses. Concentrations of oestrone, oestradiol, testosterone, 5alpha-androstane-3alpha, 17beta-diol and creatinine were measured in first morning urine samples, which had been stored since enrolment at -20 degrees C. A Cox proportional Hazards model was used, with Barlow's adjustment for case-cohort sampling, to estimate breast cancer risk in quartiles of each of the, creatinine corrected, hormone levels, the lowest quartile being the reference group. Women with higher levels of all four of the hormones were at increased risk for postmenopausal breast cancer (highest vs lowest quartile: incidence rate ratio for oestrone (IRR(oestrone)=2.5, 95% CI: 1.6-3.8; IRR(oestradiol)=1.5, 95% CI: 1.0-2.3; IRR(testosterone)=1.6, 95% CI: 1.0-2.4; IRR(5alpha-androstane-3alpha, 17beta-diol)=1.7, 95% CI: 1.1-2.7). In conclusion, women with higher excretion levels of both oestrogens and androgens have an increased risk of breast cancer.