The incidence of breast cancer was determined in 4940 women treated for tuberculosis between 1925 and 1954 in Massachusetts. Among 2573 women examined by X-ray fluoroscopy an average of 88 times during lung collapse therapy and followed for an average of 30 years, 147 breast cancers occurred in contrast to 113.6 expected [observed/expected (O/E) = 1.29; 95% confidence interval (CI) = 1.1-1.5]. No excess of breast cancer was seen among 2367 women treated by other means: 87 observed versus 100.9 expected. Increased rates for breast cancer were not apparent until about 10 to 15 years after the initial fluoroscopy examination. Excess risk then remained high throughout all intervals of follow-up, up to 50 years after first exposure. Age at exposure strongly influenced the risk of radiation-induced breast cancer with young women being at highest risk and those over age 40 being at lowest risk [relative risk (RR) = 1.06]. Mean radiation dose to the breast was estimated to be 79 cGy, and there was strong evidence for a linear relationship between dose and breast cancer risk. Allowing for a 10-year minimum latent period, the relative risk at 1 Gy was estimated as 1.61 and the absolute excess as 10.7 per 10(4) woman-years per gray. When compared to other studies, our data suggest that the breast is one of the most sensitive tissues to the carcinogenic force of radiation, that fractionated exposures are similar to single exposures of the same total dose in their ability to induce breast cancer, that risk remains high for many years after exposure, and that young women are especially vulnerable to radiation injury.