We studied differences in access to large or accredited cancer programs as a possible explanation for geographic disparities in adherence to the national guideline on lymph node assessment for Stages I to III colon cancer. State cancer registries were linked with Medicare claims of patients diagnosed from 2006 to 2008 from Appalachian counties of four states. Metropolitan and nonmetropolitan patients differed on adherence, proximity to high-volume or accredited hospitals, and hospital type. We modeled effects of hospital type on adherence with ordinary least squares and instrumental variables (instrumenting for hospital type with relative distance). The evidence was strongest for improved adherence in high-volume hospitals for nonmetropolitan patients. We estimate that roughly 100 deaths might be prevented over 5 years among each year’s incident cases if the nonmetropolitan disparity in hospital volume were eliminated nationally. We conclude that regionalization or targeting smaller hospitals would improve adherence in nonmetropolitan areas, but also argue for improving adherence generally.