I used data from the National Center for Health Statistics to map the geographic distribution of age-adjusted, race- and gender-specific stroke mortality rates in the United States from 1939-1941 to 1979-1981. Over this interval, stroke mortality rates declined dramatically with convergence of age-adjusted, state-specific stroke rates both within and between the various race-gender groups. For each race-gender group, high age-adjusted stroke mortality rates were significantly clustered in the southeastern United States, particularly in the South Atlantic census division, with persistent extreme rates in Georgia and the Carolinas. For whites, low-rate states were concentrated in the Mountain census division and along the northern Atlantic coast. The nonrandom distribution of stroke mortality across the United States, the large magnitude of the difference between high- and low-rate areas, the persistence of the pattern over more than four decades, the similarity of the distribution for different race-gender groups, the lack of delimitation by administrative or political boundaries, and results of national cooperative studies completed in the late 1960s and early 1970s together suggest that the pattern of excess stroke mortality is not an artifact of different diagnostic and reporting practices. Some of the observed geographic variation may be due to both the effects of selective migration and variations in the distributions of stroke risk factors.