Endogenous estrogens play an integral role in the etiology of breast, endometrial, and, possibly, ovarian cancers. Estrogen metabolism yields products that are potentially both estrogenic and genotoxic, yet individual metabolic patterns are just beginning to be explored in epidemiologic studies. Within the Nurses' Health Study II, we examined reproducibility of 15 urinary estrogens and estrogen metabolites (EM) among 110 premenopausal women with three luteal-phase urine samples collected over 3 years. EM were measured by a recently developed high-performance liquid chromatography-tandem mass spectrometry (LC-MS(2)) method with high sensitivity, specificity, and precision. We assessed Spearman correlations and intraclass correlation coefficients (ICC) across the three samples. Correlations between urinary estrone or estradiol and EM were only modest (r = 0.1-0.5). The 2- and 4-hydroxylation pathways were highly correlated (r = 0.9) but weakly inversely correlated with the 16-hydroxylation pathway (r = -0.2). Within-woman reproducibility over time was fairly high for the three pathways, with ICCs ranging from 0.52 (16-hydroxylation pathway) to 0.72 (2-hydroxylation pathway). ICCs were similarly high for 2-catechols and the individual catechols (ICCs = 0.58-0.72). Individual and grouped methylated 2-catechols had fairly high ICCs (0.51-0.62), but methylated 4-catechols had low ICCs (0.14-0.27). These data indicate that, in general, urinary EM levels vary substantially among individuals compared with intraindividual variability. Within-person reproducibility over time for most EM measures is comparable to or better than that for well-vetted biomarkers such as plasma cholesterol and, in postmenopausal women, estradiol.