Background: There is evidence that patients with Medicaid insurance suffer worse outcomes from surgical conditions; but there is little research about whether this reflects clustering of such patients at hospitals with worse outcomes. We assess the outcomes of patients with colon and rectal cancers at hospitals with a high proportion of Medicaid patients. Study Design: California Cancer Registry patient-level records were linked to discharge abstracts from California's Office of Statewide Health Planning and Development. All operative California Cancer Registry patients from 1998 and 1999 were included. Hospitals with > 40% Medicaid patients were labeled high Medicaid hospitals (HMH). We analyzed the odds of mortality at 30 days, 1, and 5 years for colon cancer and rectal cancer separately. Multilevel logistic regression models were constructed, using MLwiN 2.0, to include patient and hospital-level characteristics. Results: Thirty-day mortality after colon operation was worse in HMH (1% versus 0.6%; p = 0.04); as was 1-year mortality (3.4% versus 2.4%; p = 0.001). There was no substantial difference in rates of 5-year mortality. Individuals who were insured by Medicaid had worse outcomes at 5 years. Adjustment for surgical volume eliminated the effect of HMH at 30 days (1% versus 0.7%; p = 0.45) but not at 1 year (3.4% versus 2.5%; p = 0.01). Adjustment for academic affiliation did not alter these results. There were an insufficient number of rectal cancer patients to detect any differences by hospital type. Conclusions: HMH have higher postoperative colon cancer mortality rates at 30 days and 1 year but not at 5 years. The early effect can be explained by surgical volume, but additional research is needed to determine which factors contribute to differences in intermediate outcomes after operations in HMH settings.