Researchers speculate that in addition to poverty, residential patterns may contribute to black/white differences in mortality rates. But few have assessed the relationship. This study reports considerable variation in the black/white ratio of age-specific all-cause mortality rates (ages 15-24 to 65-74 years) from 1982 to 1986 among the 38 U.S. Standard Metropolitan Statistical Areas (SMSAs) with populations greater than one million in 1980. The black/white poverty-rate ratio for each SMSA was a stronger predictor of variation in the black/white mortality ratio for men than for women. For both men and women age 15-44, the level of black/white residential segregation was a significant (positive) predictor of the black/white ratio of the age-standardized death rate. This analysis also identifies SMSAs containing geographic areas with unusually high or low black/white mortality ratios, and indicates the need for more-detailed studies of explanations for such variation.