The conditional survival experience of 4502 female breast cancer cases diagnosed in Hawaii between 1960 and 1983 was studied. The calculation of conditional survival is based on patients who have already survived for a specified time period. Cox proportional hazards models were used to examine the effect of ethnicity, stage at diagnosis, and menopausal, marital, and socioeconomic statuses on conditional survival prognosis. Native Hawaiian and Filipino women showed a significantly higher risk of dying from breast cancer than any other ethnic group, followed by Chinese and Caucasian women, while Japanese women had the lowest risk. The ethnic pattern of breast cancer survival changed and ethnic survival differences decreased from 17 to 4% as time progressed. Conditional survival varied greatly with stage at first diagnosis. For localized, regional, and distant disease, the conditional survival rate increased from 92, 71, and 24% after diagnosis to 95, 85, and 65% 5 years after diagnosis, respectively, indicating that stage at diagnosis becomes less important in determining survival as time progresses. Being married and of a high socioeconomic status increased survival probability. These results suggest that considering time effect, the conditional survival rate is an informative tool to assess clinical outcome of breast cancer among ethnic groups over a long follow-up period.