The fraction of tumor cells in the DNA synthesis (S) phase of the cell cycle is a strong prognostic indicator in human breast cancer. Most studies measuring S-phase in primary tumors have used either in vitro labeling with 3H thymidine or DNA flow cytometry. The former involves a cumbersome and time-consuming radioactive assay, while the latter is heavily dependent on the quality of the preparation and the effectiveness of the algorithms used. In this study, in vivo or in vitro labeling with bromodeoxyuridine (BrdUrd) was compared with in vitro labeling using 3H thymidine in a series of 33 human breast tumors. Thymidine labeling showed strong correlation with both in vitro (r = 0.96) and in vivo (r = 0.83) BrdUrd incorporation. The reliability between observers was higher for BrdUrd counting (r = 0.94) than for thymidine counting (r = 0.87). The mean labeling index of 15 tumors labeled in vivo with BrdUrd was 5.9% compared to 4.7% for 18 tumors labeled in vitro (p = 0.34). There was poor correlation between flow cytometric and microscopic analysis of BrdUrd incorporation (r = 0.37). We conclude that microscopic analysis of in vivo or in vitro BrdUrd incorporation is a rapid and reliable method to estimate breast tumor proliferation.