OBJECTIVE: This study examines variation between hospitals in 30-day mortality after surgery for colorectal cancer (CRC) in Denmark and explores whether hospital volume and patient characteristics contribute to any variation between hospitals. BACKGROUND: Little is known about the variation between hospitals in 30-day mortality after CRC surgery, and the impact of treatment and patient characteristics that might contribute to such variation. METHODS: Hospital variation was quantified using a multilevel approach on data derived from a nationwide database of all adenocarcinomas of colon and rectum diagnosed in Denmark in 2001 to 2004. These data were linked to several central registers providing information on patient's socioeconomic status, comorbidity, and use of medication. In total 11,287 patients, who underwent surgery at any of the 43 surgical departments were included. RESULTS: Hospitals varied from 3.5% to 44.1% in 30-day mortality after emergency colon cancer surgery, and the multilevel analysis showed that emergency patients were 5 times [odd ratio (OR) = 4.6)] as likely to die within 30 days in hospitals with the worst performance compared to those with the best performance. The American Society of Anesthesiologists (ASA) score increased the variation between hospitals (OR = 5.8), whereas the other potential explanatory variables had no effect on the variation. For patients who had elective surgery for colon and rectal cancer the variation in 30-day mortality between hospitals was small and nonsignificant. CONCLUSION: Hospital variation in 30-day mortality after CRC surgery are due to differences in hospitals' ability to take care of emergency patients, especially those with high ASA scores.