This study was undertaken to evaluate the usefulness of hospital discharge data for monitoring the impact of a regional practice guideline on treatment of colorectal cancer. The aims of the study were: i) estimating process and outcome indicators; ii) exploring the relationship between patient and hospital characteristics and these indicators. Odds ratios (OR) and 95% confidence intervals (CI) were adjusted for all risk factors analysed. All 3,614 patients undergoing potentially curative resection for rectal cancer in 75 hospitals in Piedmont (Italy) between 1997 and 2000 were included. Occurrences of abdominoperineal resections (APR), postoperative complications, reinterventions and hospital mortality were 16.2%, 11.0%, 5.5% and 4.4%, respectively. APR was performed more frequently in patients with distant metastases or urgently admitted and by hospitals with a lower volume of cases (< 25 per year) compared to hospital with more than 50 cases per year (OR = 1.50, CI = 1.16-1.94). The strongest predictors for mortality and complications were: older age, distant metastases and urgency of admission. Incidence of complications and of reinterventions was also increased among males and in patients with lesions of the lower rectum. The rate of complications showed an increasing trend during the period, from 8.5% to 14.5% likely reflecting improvement in coding during time. Low hospital workload was associated to a reduced risk of complications and reinterventions, but there was evidence of underreporting of secondary diagnoses and procedures in smaller hospitals. Outcome indicator based on secondary diagnoses and procedures are of limited value in monitoring improvement of care since they reflect also differences in coding during time and among providers.