Objectives To assess the variation in risk-adjusted 30-day postoperative mortality for patients with colorectal cancer between hospital trusts within the English NHS.

Design Retrospective cross-sectional population-based study of data extracted from the National Cancer Data Repository.

Setting All providers of major colorectal cancer surgery within the English NHS.

Participants All 160 920 individuals who underwent major resection for colorectal cancer diagnosed between 1998 and 2006 in the English NHS.

Main outcome measures National patterns of 30-day postoperative mortality were examined and logistic binary regression was used to study factors associated with death within 30 days of surgery. Funnel plots were used to show variation between trusts in risk-adjusted mortality.

Results Overall 30-day mortality was 6.7% but decreased over time from 6.8% in 1998 to 5.8% in 2006. The largest reduction in mortality was seen in 2005 and 2006. Postoperative mortality increased with age (15.0% (95% CI 14.1% to 15.9%) for those aged >80 years), comorbidity (24.2% (95% CI 22.0% to 26.5%) for those with a Charlson comorbidity score ≥3), stage of disease (9.9% (95% CI 9.3% to 10.6%) for patients with Dukes' D disease), socioeconomic deprivation (7.8% (95% CI 7.2% to 8.4%) for residents of the most deprived quintile) and operative urgency (14.9% (95% CI 14.2% to 15.7%) for patients undergoing emergency resection). Risk-adjusted control charts showed that one trust had consistently significantly better outcomes and three had significantly worse outcomes than the population mean.

Conclusions Significant variation in 30-day postoperative mortality following major colorectal cancer surgery existed between NHS hospitals in England throughout the period 1998–2006. Understanding the underlying causes of this variation between surgical providers will make it possible to identify and spread best practice, improve outcomes and, ultimately, reduce 30-day postoperative mortality following colorectal cancer surgery.