PURPOSE: We sought to identify variability in surgical care for rectal cancer across the United States. In particular, we hypothesized that in large areas of the country patients are infrequently treated by proctectomy using restorative ("sphincter-sparing") techniques. METHODS: We used all-payer state hospital discharge data from 21 states to determine county level rates of restorative proctectomy vs nonrestorative proctectomy (with colostomy) for rectal cancer. County of residence data were then used to graphically represent variability in surgical care for rectal cancer. RESULTS: From January 2002 through December 2004, 19,912 proctectomies were performed for rectal cancer. Overall, restorative techniques were used in 50.1% of all patients, whereas nonrestorative techniques were used in 49.9%. In approximately one-fourth of the counties surveyed (n = 125; 26%) nonrestorative techniques were used in greater than 60% of proctectomy cases. In the majority of counties (n = 266; 54%,) nonrestorative techniques were used in 41% to 60% of proctectomy cases. Only 20.0% (n = 98) of counties were characterized by rates of nonrestorative proctectomy below 41%. The extremal quotient was 16.9, indicating significant county variability in colostomy formation for rectal cancer surgery. CONCLUSIONS: There is significant geographic variability in the rates of restorative vs nonrestorative proctectomy for rectal cancer in the United States. Large areas of the country report particularly high rates of colostomy formation after proctectomy. An in-depth population-based analysis designed to identify factors contributing to this variability in surgical treatment of rectal cancer is needed.