During the years 1945 - 1965 461 women in the city of Turku, Southwestern Finland, were diagnosed as having a biopsy-verified breast cancer. Fourhundred and thirty-nine patients (95%) with complete clinicopathologic data have now been followed up for a mean of 28 years (range from 22 to 42 years) or until death. The survival rate corrected for intercurrent deaths was 44%, 35%, and 34% 10, 20, and 30 years after the diagnosis, respectively. Only 1.2% of all deaths caused by breast cancer occurred more than 20 years after the diagnosis, and therefore about one third of the patients are likely to be cured. Fifty-six (12.8%) patients developed a second primary breast cancer or cancer of other sites. Survival of the patients diagnosed in the 1960s was better than that of the patients diagnosed earlier (p = 0.02), but the relative percentage of prognostically unfavorable poorly differentiated (Gr III) cancers became smaller with time (p = 0.009). Axillary nodal status was the most important independent prognostic factor for the 342 patients with an operable, unilateral, and invasive breast cancer in Cox's multivariate analysis (p < 0.001), followed by histologic grade, type of tumor margin, the primary tumor size (p < 0.001), and the extent of tumor necrosis (p = 0.003). Histologic type, mitotic count, nuclear pleomorphism, extent of tubule formation, amount of elastin, and extent of intraductal tumor growth were also significant prognostic factors in a univariate analysis.