In a prospective controlled clinical trial, 70 patients with normal gastrointestinal function were randomised to receive either an elemental diet based on Vivonex HN or an isonitrogenous isocalorie polymeric diet based on Clinifeed 400, administered by continuous 24 hour nasogastric infusion. The two groups of patients were well matched for age, sex, diagnosis, prior starvation, duration of feeding, initial nutritional status, and metabolic status. Nitrogen losses were significantly less on the polymeric feed, despite similar intakes. Serum transferrin rose significantly (1.85 +/- 0.2 to 2.30 +/- 0.2 g/l, p less than 0.05) only in the Clinifeed group, but nutritional parameters were otherwise maintained in both groups. The incidence of diarrhoea (Vivonex, 23.5%; Clinifeed, 30.6%) was not significantly different and was attributable to antibiotics in most cases. Hypokalaemia, which occurred in nearly half the patients, was equally distributed in the two groups, but hypophosphataemia occurred more often in the Vivonex group (p less than 0.05). Liver enzyme disturbances were similar in both groups. The present findings, therefore, provide no evidence that chemically defined 'elemental' diets containing free amino acids as their nitrogen source are in any way superior to polymeric diets containing whole protein and fat when administered to patients with normal gastrointestinal function.