To determine the most successful mode of treatment, 33 consecutive cases of duodenal atresia treated by duodenoduodenostomy and not associated with other gastro-intestinal anomalies were analysed retrospectively. These patients have been placed in a nonrandomised fashion into one of three groups: Group A: Duodenostomy (side to side) with gastrostomy and transanastomotic feeding tube (n = 12); Group B: Duodenoduodenostomy (diamond shape) with jejunostomy feeding tube (n = 12); Group C: Duodenoduodenostomy (diamond shape) only (n = 9). A nasogastric tube was used in all cases. There was no difference between the groups for gestational age, birthweight, and age at operation. The outcome measures used to compare these groups were the time taken to achieve full preanastomotic feeds and the duration of hospital stay. There was no difference in time taken to achieve full pre-anastomotic feeds between Group A and Group B. Patients in Group C took significantly less time to achieve full pre-anastomotic feeds than either of the other two groups (p < 0.05, Mann-Whitney U). The duration of hospital stay was also significantly shorter for patients in Group C (median = 12 days) than for patients in either Group A or B (median = 24, 20 days respectively) (p < 0.05, Mann-Whitney U).