Clostridium difficile colitis may be diagnosed either by endoscopy or by laboratory tests. To determine the role of endoscopy, we reviewed 59 cases of confirmed C. difficile colitis. In all patients, the etiology was confirmed by stool tests. Twenty-nine underwent lower gastrointestinal endoscopy. In 16 (55%) there was endoscopic confirmation of pseudomembranes while 4 (14%) had only nonspecific colitis. There was no apparent difference in the rate of detection of pseudomembranes between rigid sigmoidoscopy (57%), flexible sigmoidoscopy (50%), and colonoscopy (50%). Vancomycin and metronidazole were equally effective therapy but treatment with vancomycin cost more than 250 times that for metronidazole. There were no patients in whom the diagnosis was made by endoscopy alone. Endoscopy was costly and insensitive, while noninvasive stool tests were cheap and accurate. We conclude that endoscopy should be relegated to a secondary role in the workup of antibiotic-associated diarrhea.