Data from the tumor abstracts of 25,535 female breast cancer patients in California were analyzed to measure hospital differences in the rates of metastatic disease at diagnosis. For white patients the metastatic rates varied from 38 to 68 per cent as hospital affiliation varied, and for blacks they ranged from 48 to 76 per cent. Tests for normality showed that the distribution of rates for white patients in private facilities was not merely due to random variation. The metastatic rates for both racial groups tended to be inversely related to the proportion of whites among the patient populations and positively related to total patient load. Hospitals affiliated with medical schools had relatively high rates for white private patients. Several explanations of the data are suggested. They consider the possible effects of patient and physician health care practices and host vulnerability to the spread of breast cancer.