Eighty-three breasts obtained at random autopsy, which were presumed to be at low risk for development of cancer, and 107 cancer-associated breasts (containing cancer or contralateral to cancer-containing breasts), which were presumed to be at high risk, were studied in their entirety using a subgross slicer method with histologic confirmation. Twelve breasts (14 per cent) from the random autopsy series had from one to 13 radial scars each, with an average of 7.7 radial scars per involved breast; the average number of radial scars per breast for the entire series of 83 breasts from the autopsy series was 1.1. Twenty-eight breasts (26 per cent) of those from cancer-associated series had from one to 31 radial scars each, with an average of 15.5 per involved breast; the average number of radial scars per breast for the entire series of 107 cancer-associated breasts was 4.4. Radial scars are observed in breasts from patients in the same age--37 to about 85 years--in both series. Atypical epithelial hyperplasia ( epitheliosis ) was less frequent and less severe in breasts from the random autopsy series than in the cancer-associated breasts. Invasive carcinoma was not found in step sections of any of the radial scars; however, carcinoma in situ of the ductal type was found in the radiating arms of one. It is concluded that the presence of radial scars is at least a "marker" for enhanced tissue risk for cancer development, in much the same manner that some other mammary dysplastic features are "markers." The data do not support or definitely negate the hypothesis that tubular carcinoma develops in radial scars and subsequently evolves into the more common types of cancer, as proposed by some authors.