Recent studies have suggested that virtual colonoscopy (VC) and actual colonoscopy (AC) have similar efficacy for detection of polyps >6 mm. However, procedural risks with emerging technology such as VC need to be assessed before widespread implementation. We propose to demonstrate complication rates after AC that can be used for a comparative benchmark in VC. From 1994 to 1999 all patients undergoing AC who sustained perforation that required operation were analyzed for the mortality and complications. There were 26,162 consecutive colonoscopies that required 21 operations for perforation. Of these 16,948 (65%) colonoscopies were diagnostic and 9,214 (35%) were therapeutic with 11 (0.06%) and 10 (0.11%) operations respectively. Overall risk for colonoscopic perforation that requires operation was one in 1,246 (one in 1,541 for diagnostic and one in 921 for therapeutic). Five perforations were oversewn, 15 were resected (five with stoma), and one was drained. One patient died. There were two reoperations. Mortality was 0.006 per cent (one in 16,948) for diagnostic and zero for therapeutic colonoscopy. Overall risk for perforation that requires operation or mortality after AC is low. Virtual colonoscopists who propose screening and subsequent therapeutic interventions need to report high volume without complications as the perforation rate requiring operation was one in 1,246.