Time-depth breast cancer mortality data from the United States, England and Wales, Canada, and Japan, and breast cancer incidence data from Connecticut, Denmark, and Osaka, Japan, were analyzed. The temporal trends in mortality have been similar in the populations examined. Risks for individual cohorts decreased until about the cohort of 1900, when they began to increase. This pattern can account for the observed increases in the rates at perimenopausal ages (40-55 years). Although the magnitudes of mortality rates in Japan are about one fifth those in the West, they have been changing over time in much the same way as in the West. Cohort effects in incidence in Connecticut and Denmark also dipped around 1900. In Osaka, cohort effects rose from the cohort of 1901 to the cohort of 1936.