In 1983, the National Cancer Institute began a social-epidemiologic study of possible behavioral and biologic determinants of black/white racial disparities in cancer survival. The design, methodology, underlying hypotheses, and patient accrual of this study are discussed. Survival differences in four organ sites are investigated: cancers of the uterine corpus, breast, bladder, and colon. The first three sites were chosen because of significant observed black/white differentials in survival. Although racial disparities in survival from colon cancer are less prominent, this site was included because it is a leading cause of deaths attributable to cancer, because regional variations have been observed in black/white survival disparities, and because colon data permit cross-gender comparisons. Data collection centers for the study included the Georgia Center for Cancer Statistics, the Louisiana Tumor Registry, and the California Tumor Registry. Probability samples of patients newly diagnosed with these cancers were drawn from the areas served by these registries. Diagnostic years of eligibility were 1985 to 1986 for breast and colon cancer, and 1985 to 1987 for bladder and uterine corpus cancer. Data were collected by personal interview, medical records abstract, physician records, and pathology review. Analyses focus on seven main explanatory hypotheses.