The objective of this study is to develop a hypothesis about the carcinogenesis of breast cancer from a descriptive analysis. This study is an application of the age-period-cohort model (APC model) on the mortality rate of breast cancer in Taiwan over the period from 1964 to 1991. Age-specific and age-adjusted mortality rates are described and then age, period, and cohort effects are separately analysed. (Female) breast cancer mortality data in Taiwan from 1964 to 1991 were abstracted from the annual reports of the Taiwan Provincial Department of Health. The population data in Taiwan from 1964 to 1991 were abstracted from the demographic data in Taiwan compiled by the Ministry of Interior, R.O.C. The results show that both age-adjusted and age-specific mortality rates of breast cancer are increasing over the study period and with birth cohorts. The bimodal pattern becomes more apparent in later periods. The Clemmensen's hook is five years later both in the descriptive study and the APC analysis. The findings suggest a possible role of later life overnutrition as the major period effed (promoter) and a change in reproductive behavior as the cohort effect (initiator). The application of the APC model also denotes the possible role of later life overnutrition and a change in reproductive behavior after the birth cohort of 1929. These findings may help us develop hypotheses of carcinogenesis of breast cancer.