PURPOSE The efficacy of peritoneal drainage (PD) as an alternative to laparotomy (LAP) in the management of bowel perforation (PRF) in very low-birth-weight infants (VLBW < or = 1,200 g) remains uncertain. The authors hypothesized that survival of VLBW infants with PRF depends on the severity of illness rather than on the initial surgical approach. METHODS Demographic, clinical, and outcome data on all VLBW infants were abstracted prospectively over a 12(1/2)-year period. Infants with PRF were stratified by PD or by LAP. Illness acuity was compared using the sum of a 7-point scoring system based on the clinical signs determined to be of prognostic significance. The factors associated with adverse outcome and the epidemiology of PRF were also examined. RESULTS Of 937 infants, 78 with PRF required surgical intervention, consisting of PD in 32 (41%) and LAP in 46 (59%). Mean birth weight, illness acuity score, and the number of infants with NEC were significantly lower in PD (P =.0005). A higher proportion of PD infants received indomethacin (P =.01). There were no other differences between the 2 groups. Regardless of the choice of procedure, birth weight did not affect mortality rate; however, a shorter interval between PRF identification and surgical intervention was associated with improved survival rate (P =.001). Postoperative liver dysfunction, short gut syndrome, and enteric stricture were more common among LAP. Mortality rate, however, did not differ. When severe thrombocytopenia (P <.03) or neutropenia was present (P <.03), outcome of LAP was better than PD. Rescue LAP for 8 of rapidly deteriorating PD infants saved 5. Regardless of surgical approach, coagulopathy (P <.003), severe thrombocytopenia (P <.005), neutropenia (P <.0001), and multiple organ failure (P <.0001) were all predictive of fatality. CONCLUSIONS Choice of surgical approach should be based on the underlying illness and not on birth weight. In the presence of clinical indication of necrotic gut, or profound abdominal infection, LAP is a better choice. PD, however, is far less morbid and should be considered for isolated PRF. Rescue LAP must be considered without delay when PD fails.