The incidence and survival rates for breast cancer among residents of Rochester, Minnesota, from 1975 to 1982 were compared with rates from 1935 to 1974. The age-adjusted incidence rates of breast cancer, after exclusion of patients with carcinoma in situ, increased 14% between the periods 1965 to 1974 and 1975 to 1982 (from 87.2 to 99.5 per 100,000 person-years). Much of this increase was due to the greater number of patients with less advanced disease: the frequency of both regional lymph node involvement and distant metastatic disease at initial diagnosis decreased. In addition, the incidence of carcinoma in situ more than doubled between the periods 1935 to 1944 and 1975 to 1982. These increased proportions of less advanced disease coincide with increased public and physician awareness of the importance of early detection. No change was demonstrated in overall survival during the study period, perhaps because the follow-up for the final study period was shorter than that for the previous periods. When divided into subsets by staging characteristics, only patients with distant metastatic disease had statistically significant improvement in survival over time (P less than or equal to 0.01)--from a median survival of 21 months in the period 1955 to 1974 to 28 months in the period 1975 to 1982. This increased survival is probably related to advances in therapy (such as combination chemotherapy). In addition, earlier detection of distant metastatic disease in the later study periods might have produced the apparent improvement in survival in this subgroup.