Blacks and Whites have very different cancer incidence rates for many sites, but this may largely be due to the racial differences in socioeconomic status (SES). The authors tested this hypothesis by determining the effect of adjustment for SES on the black/white incidence ratios for 12 cancer sites. Race-specific census tract-level SES variables (median family income, percent below poverty level, and years of education) were obtained from the 1980 US census and applied to approximately 20,000 black and 88,000 white cancer cases from the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) program for the years 1978 to 1982. For each cancer site (with each sex considered separately), Poisson regression was used to produce age-adjusted black/white incidence ratios, with and without adjustment for SES. The SES variable with the strongest adjusting power was percent below poverty level. For many sites (breast, in situ and invasive cervix, esophagus, male lung, pancreas, stomach) poverty accounted for much or all of the racial differences. For several sites (bladder, multiple myeloma, prostate, uterine corpus), large racial differences persisted after adjustment for poverty, and these findings suggest directions for investigating the etiology of these cancers.