During a 17-month period (01/11/85-05/31/86) 225 cases of nosocomial diarrhea were identified in a children's hospital. Diarrhea was considered to be nosocomial if it began at least 72 hours after the patient's hospital admission or within 3 days after discharge. One or more routine diagnostic studies for identification of a pathogen were performed in 195 (87%) cases. The most commonly performed test was the bacterial stool culture. None of these samples yielded a bacterial pathogen. The only pathogens detected by routine laboratory studies were rotavirus (61/137 [45%] samples were positive for rotavirus by ELISA) and Clostridium difficile (9/54 [17%] positive for toxin). Of the patients whose tests were positive for rotavirus 56 were younger than 2 years of age, and all were identified in the winter and spring. When multiple stool samples were tested by the diagnostic laboratory, rotavirus was identified in an additional 14 patients whose initial stool samples were negative for rotavirus. All patients whose tests were positive for C. difficile toxin had received antibiotics within the previous 3 months. Ova/parasites were not detected in 53 of the tested stools. We also identified enteric adenovirus in six patients. Viruses were identified in 95 (42%) of the 225 cases of nosocomial gastroenteritis. Nosocomial diarrhea is common in a children's hospital. Rotavirus is the most commonly identified pathogen. Rotavirus testing is valuable in children with nosocomial diarrhea who are younger than 2 years of age, especially in the winter and spring. Multiple samples may be necessary to identify rotavirus. C. difficile toxin assay should be considered for patients who are receiving or who have received antibiotics.(ABSTRACT TRUNCATED AT 250 WORDS)