Résumé : Aim:  Anastomotic leakage (AL) after total mesorectal excision (TME) is a major adverse event. This study evaluates variability in AL between centres participating on a voluntary basis in PROCARE, a Belgian improvement project, and how further improvement of the AL rate might be achieved. Method:  Between January 2006 and March 2011, detailed data on 1815 patients, (mean age 65.5 years, 63% male), who underwent elective TME with colo-anal reconstruction for rectal cancer were registered by 48 centres. Variability in early clinical AL rate was analysed before and after adjustment for gender, age >60 years, ASA score 3 or more and BMI >25 kg/m(2) . Results:  The overall AL rate was 6.7% (95% CI: 5.6% - 7.9%). Early AL required re-operation in 86.8% of patients. It increased length of hospital stay from 14.7 days to 32.4 days and in-hospital mortality from 1.1% to 4.8%. Statistically significant variability in AL rate between centres was not observed, neither before nor after risk adjustment. Nonetheless, further improvement may be achievable in some centres by targeting the adjusted performance of better performing centres. These centres used neoadjuvant treatment, rectal irrigation, mobilisation of the splenic flexure, resection of the sigmoid colon, side-to-end colo-anastomosis with or without pouch and defunctioning stoma at primary surgery in a significantly higher proportion of patients than less well performing centres. Conclusion:  The overall AL rate was low but needs to be interpreted with caution because of incomplete registration. Further improvement might be achieved by adopting the approach of better performing centres.