Effective methods of serial epithelial sampling to measure breast-specific biomarkers will aid the rapid evaluation of new preventive interventions. We report here a proof-of-principle phase 2 study to assess the utility of ductal lavage (DL) to measure biomarkers of tamoxifen action. We enrolled women with a 5-year breast cancer risk estimate >1.6% or the unaffected breast of women with T1a or T1b breast cancer. After entry DL, participants chose tamoxifen or observation and underwent repeat DL 6 months later. Samples were processed for cytology and immunohistochemistry for estrogen receptor alpha, Ki-67, and cyclooxygenase-2. Of 182 women recruited, 115 (63%) underwent entry and repeat DL; 85 (47%) had sufficient cells for analysis from > or =1 duct at both time points; in 78 (43%), cells were sufficient from > or =1 matched ducts. Forty-six women chose observation and 39 chose tamoxifen. We observed greater reductions in the tamoxifen group than in the observation group for Ki-67 (adjusted P = 0.03) and estrogen receptor alpha (adjusted P = 0.07), but not in cyclooxygenase-2 (adjusted P = 0.4) labeling. Cytologic findings showed a trend toward improvement in the tamoxifen group compared with the observation group. Interobserver variability for cytologic diagnosis between two observers showed good agreement (kappa = 0.44). Using DL, we observed the expected changes in tamoxifen-related biomarkers; however, poor reproducibility of biomarkers in the observation group, the 53% attrition rate of subjects from recruitment to biomarker analyses, and the expense of DL are significant barriers to the use of this procedure for biomarker assessment over time.