Site-specific studies of tumor incidence among Hiroshima and Nagasaki A-bomb survivors (120,000 subjects) indicate that, for some tissues, excess radiation-related risk may be confined to a relatively few cell types. Risk of malignant and benign salivary gland tumors increased strongly with radiation dose, but was largely confined to two relatively nonfatal histological subtypes, mucoepidermoid carcinoma and Warthin's tumor. There was no evidence that radiation susceptibility varied by sex or by age at exposure. Analyses of skin cancer incidence data found a statistically significant dose response for basal cell carcinoma, but not for squamous cell carcinoma. The basal cell cancer response was strongly modified by age at exposure with the highest levels following childhood exposure, and there was little evidence of an interaction with ultraviolet radiation exposure from sunlight. Preliminary analyses of incidence data for benign and malignant tumors of the central nervous system (CNS) found a highly significant radiation-related excess risk of neurilemmoma, and a smaller, but still significant, excess for a combination of all other CNS tumors combined. The latter excess appeared to involve pituitary tumors, meningiomas, and a mixture of other or """"""""not otherwise specified"""""""" tumors, but not gliomas. New data on female breast cancer are informative about risk following exposure before 20 years of age. There is no evidence that any age interval in that range is more or less sensitive to radiation carcinogenesis than any other. Risk of early-onset breast cancer (i.e., before age 35) was highly dose dependent, and epidemiological and molecular investigations are in progress to investigate the possible influence of a genetically susceptible population subgroup. Risk in later life was markedly lower (relative risk at 1 Sv = 3.3 vs. 15 for early-onset cancer), but did not vary significantly between 35 and 65 years of age.