To determine whether surgery could be avoided in some patients with perforated peptic ulcer, we conducted a prospective randomized trial comparing the outcome of nonoperative treatment with that of emergency surgery in patients with a clinical diagnosis of perforated peptic ulcer. Of the 83 patients entered in the study over a 13-month period, 40 were randomly assigned to conservative treatment, which consisted of resuscitation with intravenous fluids, institution of nasogastric suction, and intravenous administration of antibiotics (cefuroxime, ampicillin, and metronidazole) and ranitidine. Eleven of these patients (28 percent) had no clinical improvement after 12 hours and required an operation. Two of the 11 had a perforated gastric carcinoma, and 1 had a perforated sigmoid carcinoma. The other 43 patients were assigned to immediate laparotomy and repair of the perforation. One of these patients was found to have a perforated gastric carcinoma. The overall mortality rates in the two groups were similar (two deaths in each, 5 percent), and did not differ significantly in the morbidity (infection, cardiac failure, or renal failure) rates (40 percent in the surgical group and 50 percent in the nonsurgical group). The hospital stay was 35 percent longer in the group treated conservatively. Patients over 70 years old were less likely to respond to conservative treatment than younger patients (P less than 0.05). We conclude that in patients with perforated peptic ulcer, an initial period of nonoperative treatment with careful observation may be safely allowed except in patients over 70 years old, and that the use of such an observation period can obviate the need for emergency surgery in more than 70 percent of patients.