OBJECTIVE: To evaluate the risk factors associated with a nosocomial outbreak of Clostridium difficile-associated diarrhea. DESIGN: Case-control study with two control groups. SETTING: University-affiliated urban hospital. PATIENTS: A convenience sample of 26 patients was chosen out of a total of 78 hospitalized patients with C difficile-associated diarrhea, defined as the presence of diarrhea and a positive C difficile cytotoxin assay or stool culture. Twenty-six controls were matched for age, gender, ward, and date of admission; 18 additional controls were matched to surgical patients for date and ward of admission, as well as for the type of surgical procedure performed. RESULTS: Significant risk factors for the development of C difficile-associated diarrhea were gastrointestinal surgery (exposure odds ratio [EOR] = 7.9, p = .004, 95% confidence interval [CI] = 1.9, 35), use of neomycin (EOR = 15.6, p = .012, 95% CI = 1.7, 92), clindamycin (EOR = 15.6, p = .005, 95% CI = 1.7, 92), metronidazole (EOR = 5.7, p = .02, 95% CI = 1.4, 25), and excess antibiotic use (mean number of antibiotics = 4.2 versus 1.4, p < .00005). The presence of systemic disease and the use of antacids or immunosuppressive drugs were similar in cases and controls. In surgical patients, no specific antibiotic could be linked to C difficile-associated diarrhea because of uniform perioperative antibiotic use. There was a significant difference in the number of antibiotics administered to cases and controls (mean = 3.1 versus 1.9, p < .005). CONCLUSIONS: The results suggest that control of nosocomial C difficile-associated diarrhea may be attained by minimizing the administration of antibiotics, avoidance of high-risk antibiotics, and having a high index of suspicion of C difficile-associated diarrhea in patients who develop diarrhea after gastrointestinal surgery. Perioperative administration of metronidazole, when given with other antibiotics, failed to protect against the development of C difficile-associated diarrhea.