There were nearly 400 cases of gastro-enteritis in Brisbane from May to October 1947. About half represented the endemic level then prevailing, and the remainder formed a virulent epidemic, which was responsible for most of the deaths.Salmonella bovismorbificanswas isolated from a sufficient proportion of these cases to justify the belief that it was the cause of the epidemic.Of recognized infections withS. bovis-morbificans77 % were in babies under 1 year old, 19 % in those 1–2 years old, and 4 % in older children and adults. The infection was most severe in the under-1-year-old group, and severity varied during the epidemic, reaching a maximum in August and thereafter declining steadily.Nearly two-thirds of the infections were institutional.Infants were found to be infective during the incubation period and also for periods up to 6 months after clinical recovery. Their faeces provided the greatest mass of infective material. Infections in older age groups were relatively unimportant in relation to the rate of epidemic spread, but significant secondary centres of infection developed in mice and cockroaches in the hospital wards.The path of infection was traced from the faeces to the sinks and brushes used for cleansing soiled napkins, to mice and cockroaches which had access to the sinks, and to ward wash-basins and nailbrushes used by the nursing staff. Possible indicators of faecal contamination were found on nurses' hands, in ward kitchens, and in milk mixtures prepared for babies' feeds.Control is believed to be practicable and to depend, first, on making medical officers and nurses aware of the dangers; and secondly, on improved ward hygiene, of which elimination of washing napkins in wards, reduction of contamination when changing napkins, rigid segregation of ‘clean’ and ‘dirty’ nurses, and improvements in food preparation and administration, are regarded as the most important.