The persistent diarrhoea-malnutrition syndrome (PDM) remains a leading cause of morbidity and mortality in hospitals in resource-poor countries. In view of the benefits of elemental or oligomeric feeds in inflammatory bowel diseases, we performed a randomized controlled trial of an exclusive diet of amino acid-based elemental feed (AAF) compared with standard nutritional rehabilitation (based on skimmed milk and then soya) for PDM. Treatment was given for 4 weeks in the malnutrition ward of the University Teaching Hospital, Lusaka, in a single-blind study. Intestinal and systemic infections were treated with routine therapies. The main outcome measures were weight gain, recovery from diarrhoea, and mortality. Two hundred children (106 HIV seropositive, 90 HIV seronegative) were randomized; 155 children completed therapy, 39 died, and six were lost to follow-up. At randomization, they were severely malnourished: median baseline weight-for-age z-score was -4.0 (interquartile range, IQR -4.4, -3.5); 9 per cent were underweight, 23 per cent had marasmus, 47 per cent had kwashiorkor, and 21 per cent had marasmic-kwashiorkor. Weight gain was greater in the AAF group (median gain in weight-for-age z-score was 1.23, IQR 0.89-1.57) compared with the control group (0.87, IQR 0.47-1.25; p=0.002), although calorie intakes were higher in the control group. The increase in haemoglobin concentration was also greater in the AAF group (0.8 g/dl, IQR 0-1.8) than in the control group (0.3, IQR -0.6, -1.6; p=0.04). Diarrhoea frequency and global recovery scores improved equally in both treatment groups and mortality did not differ. A diet of reduced molecular complexity was associated with significantly improved weight gain.