Greater use of mammography in the United States in recent years has increased the detection of early neoplasms of the breast, including carcinoma in situ. However, the occurrence and treatment of diagnosed carcinoma in situ of the breast has not been fully described. Our goal was to examine temporal, geographic, and demographic patterns in the incidence and treatment of in situ breast cancer. The study included data from all women with in situ breast cancer that had been detected in the nine Surveillance, Epidemiology, and End Results areas of the United States from 1975 through 1990 (Surveillance Program, Cancer Statistics Branch, Bethesda, MD: National Cancer Institute, November, 1993). We calculated age-adjusted incidence rates (1970 United States standard) using data on histology and treatment from the Surveillance, Epidemiology, and End Results data tape. We assessed predictors of treatment by mastectomy using multiple logistic regression. From 1975-1979 to 1986-1990, the age-adjusted incidence rate of in situ breast cancer increased from 4.7 to 16.9/100,000 women. The increase occurred in all age groups and among both white and black women. However, there was nearly a 2-fold difference in incidence rates across geographic areas in 1986-1990, ranging from < 12/100,000 in Iowa and New Mexico to > 20/100,000 in San Francisco and Seattle. Geographic variability in treatment was also evident, with mastectomy, rather than breast-conserving therapy, performed on 46% of the women with in situ breast cancer in San Francisco and on 66% of those in Iowa. The incidence of diagnosed in situ breast cancer increased markedly during the 1980s, and there was substantial geographic variability in the rates of detection of these tumors and in the type of therapy received. Although mastectomy became a less common treatment over time, it was still performed on a high proportion of women with in situ breast cancer during the latter part of the decade.