Levels of total estradiol in premenopausal women vary widely over the course of the menstrual cycle with a spike at the time of ovulation and dissimilar patterns pre- and post-ovulation. Evaluating the association between breast cancer and premenopausal measurements of total estradiol when the measurements cannot be taken on a uniform day of the cycle is therefore a difficult methodological challenge. In a matched case-control study of breast cancer nested within a prospective study, premenopausal serum samples obtained up to 7 years before breast cancer diagnosis were available for total estradiol assay. By fitting a three-piece spline model that regressed the logarithm of total estradiol (ln estradiol) on day of menstrual cycle, the authors were able to adjust the measurements for day of the cycle on which they were collected by expressing them in terms of the number of standard deviations above or below the fitted ln estradiol value for that day. Applying the adjusted measurements to the nested case-control study, they found evidence of a 1.5 to 2-fold risk for women in the upper two tertiles of ln estradiol relative to women in the lowest tertile. Conditional logistic regression analysis for day-of-cycle-adjusted ln estradiol treated as a continuous variable resulted in a relative risk estimate of 1.19 (95% confidence interval 0.91-1.55) per standard-deviation increase in adjusted ln estradiol.