There is considerable controversy regarding the role of estrogen metabolites in breast cancer risk, fueled in part by the development of a rapid ELISA that is suitable for large scale investigations. An earlier version of the ELISA could detect values of the 2-hydroxyestrone (2-OHE1) and 16alpha-hydroxyestrone (16alpha-OHE1) metabolites as low as 2 ng/ml and produce consistent results in premenopausal urines. However, reproducibility was problematic in postmenopausal urines where concentrations of these compounds are much lower. In response to our concern, a new ELISA was developed with a sensitivity of 0.625 ng/ml, which we evaluated using the same pre- and postmenopausal urine samples analyzed in the earlier ELISA. In this report, we present findings on the new kit with regard to reproducibility of the 2-OHE1 and 16alpha-OHE1 measurements, comparability of results with gas chromatography-mass spectroscopy values, and with regard to the stability of the metabolites after repeated freeze-thaw cycles and after preservation by boric acid. For the most part, we found the new ELISA to be reproducible, with assay coefficients of variation ranging from 10 to 20%, and intraclass correlation coefficients (ICCs) ranging from 80 to 95% in both the pre- and postmenopausal urines. ELISA results for 16alpha-OHE1 differed from 1 day (i.e., batch) to the next, and the absolute values of the metabolites obtained by the ELISA were consistently lower than but well correlated with those obtained by gas chromatography-mass spectroscopy. Values of the 2-OHE1:16alpha-OHE1 ratio also differed between the methods, but because the range of values was not large, the magnitude of these differences was not as great. For the ratio, the correlation between methods was excellent, and the ICCs were high for both groups of women. After preservation by boric acid, values of the ratio varied according to acid concentration but not in a linear fashion. Ratio values were similar in urine samples exposed to four different freeze-thaw cycle treatments, although values for all treatments were consistently lower in one batch. Because batch-to-batch variability was not negligible, it is advisable that matched cases and controls be analyzed in the same batch. Provided this is done, the relatively low assay coefficient of variation and high ICC demonstrate that the new ELISA kit can reliably measure the 2-OHE1:16alpha-OHE1 ratio and detect small case-control differences in large population-based studies, where rapid and relatively easy laboratory methods are critical.