BACKGROUND: Access to health insurance influences the amount and quality of health care received, which in turn is likely to be related to survival. Few studies have systematically examined cancer survival by individual level health insurance data from a state population-based cancer registry for 4 anatomic sites. METHODS: Men and women aged 18 to 99 years who were registered from 1995 to 1998 with the Kentucky Cancer Registry, Lexington, with colorectal, lung, breast, or prostate cancer were followed up through 1999. Three-year crude and relative survival proportion by 7 health insurance categories and by sex for all 4 sites were calculated. Poisson regression was used to model the risk of death (controlling for age group at diagnosis, sex, race, stage at diagnosis, and treatment) relative to expected deaths in the general population from all 4 cancers by health insurance category. RESULTS: Among patients with prostate cancer, 3-year relative survival proportion was 98% for the privately insured and 83% for the uninsured; comparable figures were 91% and 78% for patients with breast cancer; 71% and 53% for patients with colorectal cancer; and 23% and 13% for patients with lung cancer. For all 4 cancers the uninsured ranked fifth or sixth on survival, above patients with unknown insurance type or Medicaid/welfare. CONCLUSION: These findings confirm purported disparities in cancer care and point toward the need to make quality care accessible to all segments of the population.