BACKGROUND The aim of the study was to determine whether the detection of Clostridium difficile toxin in stools may be more profitable than conventional stool cultures for the etiologic study of nosocomial diarrhea and to analyze what risk factors favor the development of nosocomial diarrhea by C. difficile. METHODS The presence of enteropathogens and A and B toxins of C. difficile were investigated (by monoclonal antibody enzymoimmunoassay) in stools of patients with nosocomial diarrhea. A series of patients simultaneously admitted without diarrhea were selected as the control group. RESULTS During a 6 month period 92 patients with nosocomial diarrhea and 82 controls without diarrhea were studied. The C. difficile toxin was detected in 8 of these 174 patients (4.6%). Eight point seven percent of the nosocomial diarrheas were related with C. difficile while only 1% were due to an enteropathogen (Salmonella enteritidis). C. difficile toxin was not detected in any patient who did not have diarrhea. In comparison with the patients with diarrhea due to other causes, the patients with diarrhea by C. difficile had more frequently received antibiotics over the previous 7 days (57 vs 88%) and had been hospitalized for a longer time (> or = 7 days) (58 vs 88%) (p < 0.05). CONCLUSIONS In the author's institution infection by Clostridium difficile is the most frequent cause of nosocomial infectious diarrhea, especially in patients admitted for a prolonged time or who receive antibiotics. The routine investigation of enteropathogens in the cases of nosocomial diarrhea does not seem justified while the detection of the A and B toxins of C. difficile may be more profitable.