INTRODUCTION: Small intestinal bacterial overgrowth refers to the proliferation of abnormal numbers and types of microorganisms in the lumen of the proximal bowel. Bacterial overgrowth has been implicated as a possible factor in prolonging some episodes of infantile gastroenteritis. This thesis examines 2 different aspects of the duodenal flora of infants with gastroenteritis, and has therefore been divided into 2 separate studies. CARBOHYDRATE STUDY: Objective: To test the hypothesis that during a diarrhoeal episode the presence of malabsorbed carbohydrate in the duodenal lumen acts as a factor promoting bacterial proliferation. Patients and methods: Infants admitted to the rehydration ward with acute gastroenteritis were selected for study if they fulfilled various criteria in terms of age, nutritional status, previous diarrhoeal episodes and antibiotic administration. They were admitted to the research ward. Weights were measured and if they had severe diarrhoea (≥ 30g/kg) were included in the study. Twenty patients were entered into the study. On admission into the trial the first duodenal intubation was done to measure the duodenal flora quantitatively and qualitatively. Thereafter the patients were assigned on an alternate basis to one of 2 groups. One group (carbohydrate-containing group) received a soy-based infant formula containing carbohydrates (Isomil, Ross). The other group (carbohydrate-free group) received an identical milk but from which all carbohydrate had been omitted (Ross CHO-free). To these infants carbohydrate was given intravenously. Stool output was measured daily. After 3 days of the respective diets the duodenal flora was re-examined. Results: Longitudinal analysis of the duodenal flora of the carbohydrate-containing group showed a small decrease in the number of bacterial isolates and in their magnitude. The duodenal flora of the carbohydrate-free group was virtually unchanged. Comparing the duodenal bacteriology of the groups the only significant difference was that the number of isolates and the magnitude of Haemophilus was greater in the carbohydrate-free group- (p < 0.05). The diarrhoea resolved in 5 patients: 2 in the carbohydrate-containing and 3 in the carbohydrate-free group. Conclusions: The lack of difference in the response of the duodenal flora between the two groups studied suggests that the presence of carbohydrates in the lumen is not important in encouraging the growth of bacteria in that site. The possible causes for an increase in Haemophilus numbers in the carbohydrate-free group are discussed. BOWEL COCKTAIL STUDY: Objective: Small intestinal bacterial overgrowth has been proposed as a cause of progression of acute diarrhoeal episodes to persistence. The "bowel cocktail", a combination of oral gentamicin and cholestyramine, has been shown to be effective in terminating episodes of persistent diarrhoea. It has been postulated to work by eradicating small intestinal bacterial overgrowth, but its mode of action is not known. The objective of this study was to examine the changes in the duodenal flora associated with administration of the bowel cocktail in order to elucidate its possible mechanism or mechanisms of action. Patients and methods: The study group comprised 15 patients. Fourteen were infants from the carbohydrate study who had ongoing diarrhoea. The remaining infant (the "late entry") was selected from the rehydration ward. Severe diarrhoea, as defined by a stool output equal to or greater than 30g/kg/day, was a pre-requisite for entry into the study. The investigation involved 2 duodenal intubations for microbiological analysis of the duodenal fluid. After the first intubation (which was the second intubation for the 14 infants who had been in the carbohydrate study) the bowel cocktail was administered. This comprised a 3-day course of oral gentamicin and 5 days of oral cholestyramine. Forty-eight hours after the start of therapy the duodenal bacteriology was repeated. The patient management was the same as during the carbohydrate study and the feeding regimen of the infants was not altered. The study ended immediately after completion of the bowel cocktail course. Results: Administration of the bowel cocktail was associated with a decreased stool output in all patients. Bacteriological analysis of the duodenal flora after this treatment showed a statistically significant decrease in the total microbial count, the aerobic microbial fraction and the Enterobacteriaceal fraction. On analysis of the bacterial genera a significant decrease was noted in Neisseria, Haemophilus, and aerobic lactobacilli. Analysis of individual patients' duodenal fluid bacteriology in conjunction with the stool bacteriology results before administration of the bowel cocktail often provided an explanation as to the possible aetiology of the diarrhoea and its resolution by therapy. Conclusions: Small intestinal bacterial overgrowth, in the accepted sense of a luxuriant flora teeming with faecal organisms, did not appear to be a feature of the patients in this study. The total bacterial count was only slightly above the accepted upper limit of normal. Although the decrease in the number of Enterobacteriaceae could possibly be interpreted in the context of bacterial overgrowth, a study of the individual patients' duodenal flora shows that these microorganisms were more likely to be acting as specific enteric pathogens. It is concluded that small intestinal bacterial overgrowth, as currently defined, is not an important cause of persistent diarrhoea. The efficacy of the bowel cocktail is more likely to reside in its ability to eradicate specific enteric pathogens. The author ends by questioning the validity of the whole concept of small intestinal bacterial overgrowth.