INTRODUCTION: Approximately 145,000 Americans were diagnosed with colorectal cancer and 56,000 died from colorectal cancer in 2006. Although colorectal screening can reduce mortality and incidence, U.S. screening rates are particularly low for racial/ethnic minorities. Racial differences in the subsite location of colon cancers could have implications for colorectal screening. This study examines the anatomic subsite distribution of tumors among African-American, Hispanic, Asian-American/Pacific-Islander and non-Hispanic white (NHW) patients with colon cancer. METHODS: Surveillance and End Results program data for 254,469 primary in situ and invasive colon cancers for patients from 1973-2002 are included in this analysis. Descriptive analyses and logistic regression are used to describe and examine variations in the proportion of colon cancers diagnosed at sites proximal to the sigmoid colon or proximal to the splenic flexure over three successive time periods. RESULTS: The proportion of colon cancers diagnosed at the sigmoid colon was 15.6-21.3% lower, while diagnoses at the descending colon were 40.5.0-45.3.0% higher for African Americans than NHWs over the three time periods. In logistic regression analyses adjusted for gender, age group and year of diagnosis, the odds of a diagnosis of cancer proximal to the sigmoid colon or proximal to the splenic flexure was significantly higher for African Americans but lower for Hispanics and Asian Americans/Pacific Islanders compared to NHWs. DISCUSSION: The higher proportion of cancers among African Americans diagnosed at sites that are generally attempted but not always reached with a sigmoidscope suggest that African Americans may benefit from screening colonoscopy. They also highlight the need for systems that collect data that would allow a direct examination of the role that the differential use of specific colon screening tests and polypectomy play in racial/ethnic variation in colon cancer incidence and in the anatomic subsite distribution of colon cancers.