This study assessed the relationship of race and patterns of care, defined by an expert NCI-appointed committee, for 7,781 patients with breast cancer treated in 107 hospitals in 45 communities between 1982 and 1985. After control for age and stage of disease, black patients had significantly different care from white patients for four of the ten patterns examined. They were less likely to have a progesterone receptor assay or to be referred for postmastectomy rehabilitation, two patterns deemed desirable for all patients. Black patients were also more likely to receive liver scans and radiation therapy in situations in which these procedures were labeled "less appropriate (as defined in the text)." Black patients differed significantly from whites on their health insurance, hospital, and physician characteristics; these factors were also significantly associated with the patterns of care. However, after controlling for these variables, the association between race and care persisted for three patterns. The patterns that showed racial differences were not the most clinically significant of the ten studied. Different treatment for black and white patients may help to explain differences in survival rates of black and white women with breast cancer.