The aim of this study was to determine by univariate and multivariate analyses the factors associated with clinically significant anastomotic leakage (AL) after large bowel resection. From 1990 to 1997 a series of 707 patients underwent colonic or rectal resection (without a stoma). Patients were divided into two groups: those with clinical anastomotic leakage (group 1) and those without it (group 2). AL occurred in 43 of 707 patients (6%). The overall mortality was 2.2% and was significantly higher in patients with AL than in those without: 5 of 43 (12%) versus 11 of 664 (1.6%), p < 0.001. Univariate analysis showed 15 variables associated with the risk of AL: previous abdominal or pelvic irradiation (p = 0.02), American Society of Anesthesiologists (ASA) score > 2 (p = 0.04), leukocytosis (p = 0.02), renal failure (p = 0.03), steroid treatment (p = 0.01), duration of operation (p = 0.001), intraoperative septic conditions (p = 0.006), total colectomy (p = 0.009), transverse colectomy (p = 0.02), difficulties encountered during anastomosis (p = 0.001), ileorectal anastomosis (p = 0.02), colocolic anastomosis (p = 0.01), abdominal drainage (p = 0.05), and blood transfusion intraoperatively (p = 0.006) and postoperatively (p = 0.001). Multivariate analysis showed that only preoperative leukocytosis (p = 0.04), intraoperative septic conditions (p = 0.001), difficulties encountered during anastomosis (p = 0.007), colocolic anastomosis (p = 0.004), and postoperative blood transfusion (p = 0.0007) were independent factors associated with AL. The risk of AL increased from a range of 12% to 30% if one risk factor was present, to 38% with two factors, to 50% with three factors. After colorectal resection and intraperitoneal anastomosis, a temporary protective stoma is proposed in selected patients with high risk factors for AL, as observed in our study.