Anastomotic leakage is a major problem in colorectal surgery, and previous studies have suggested that intraoperative identification of leaks allows repair at the time of surgery. This study examined whether testing allowed a defunctioning ileostomy to be safely omitted. A series of 102 consecutive patients underwent left-sided colorectal resection, 52 males and 50 females, mean age 65.7 years (range 16-89 years). After completion of the anastomosis, its integrity was tested by running saline into the rectum, using a manometer, to a maximum distending pressure of 30 cmH2O. Any leaks were repaired and the anastomosis retested. A defunctioning ileostomy was only performed if the anastomosis could not be shown to be leak-proof on testing. Patients underwent a contrast enema on the 8th postoperative day. Twenty-one (20.6%) patients failed the initial leakage test and 3 (3%) patients failed a second test. Two of these 21 patients went on to have a clinical leak, both of which were treated conservatively. Two defunctioning ileostomies were performed at the time of surgery. Sixteen (16.2%) had a leak on radiological testing, and there was clinical evidence of a leak in 5 (4.9%) patients. There were 3 (2.9%) deaths, but none of these had a leak on radiological testing. Incomplete anastomoses were successfully corrected intraoperatively. A defunctioning ileostomy was avoided in 98% of cases. Intraoperative testing to a pressure of 30 cmH2O is helpful in anterior resection, but does not guarantee that an intact anastomosis will remain intact postoperatively.