To assess the usefulness of routinely collected socioeconomic variables from the U.S. census in predicting burn incidence rates, burn rates and 25 socioeconomic variables were analyzed at the level of census tracts for the Boston Standard Metropolitan Statistical Area. The burn rates were based on data collected during the National Burn Demonstration Project and consisted of patients who sustained burns between July 1, 1978, and June 30, 1979, and who required inhospital care. Analysis of the data revealed strong associations between burn rates and six of the variables. The six variables were the percentage of families below the poverty level with a householder under age 65 years, the percentage of persons over 5 years of age whose residence in 1975 was a different dwelling but in the same county, the percentage of persons in the civilian labor force who were unemployed at the time of census enumeration, the average age of occupied dwelling units, the percentage of occupied housing units occupied on a rental basis, and the percentage of persons age 25 years or older who acquired some college education but did not complete college. Interpretation of the findings is not straightforward but seems to suggest that the previously observed association between poor socioeconomic status and increased burn risk for individuals can be quantified at the census tract population level.