Significant production of breath hydrogen has been shown in premature infants, suggesting limited intestinal capacity for digestion of carbohydrate. To evaluate net absorption of carbohydrate 24 three day balance studies were carried out in seven preterm infants fed pasteurised banked human milk and in 17 preterm infants fed a formula containing 75% lactose and 25% glucose polymers. Because carbohydrate reaching the colon may be converted to organic acids by bacterial flora, carbohydrate net absorption was determined by quantitating the faecal excretion of energy derived from carbohydrate. The carbohydrate derived energy content of milk and stools was calculated as the difference between the measured gross energy and the sum of energy related to nitrogen and fat. Faecal loss of carbohydrate derived energy was lower in the group fed formula (1.9 (SD 1.2) kcal/kg/day) than in the group fed human milk (4.0 (SD 1.8) kcal/kg/day). Net absorption of carbohydrate derived energy was 97.0 (SD 1.9)% as opposed to 92.6 (SD 3.9)%, respectively. Within each group there was no significant relation between carbohydrate energy absorption and fat, nitrogen, or gross energy absorption. Thus, although less complete with human milk than with formula, apparent absorption of energy derived from carbohydrate seemed quite satisfactory in these preterm infants.