Controversy remains as to which patients with small bowel obstruction (SBO) need immediate surgery and which may be managed conservatively. This study evaluated the ability of clinical risk factors to predict the failure of nonoperative management of SBO. The electronic medical record was used to identify all patients with SBO over one year. Clinical, laboratory, and imaging data were recorded. Univariate and multivariable analyses were performed to identify risk factors predicting need for surgery. Cox proportional hazards regression was used to identify risk factors that influence need and timing for surgery. Two hundred nineteen consecutive patients were included. Most patients did not have a prior history of SBO (75%), radiation therapy (92%), or cancer (70%). The majority had undergone previous abdominal or pelvic surgery (82%). Thirty-five per cent of patients ultimately underwent laparotomy. Univariate analysis showed that persistent abdominal pain, abdominal distention, nausea and vomiting, guarding, obstipation, elevated white blood cell count, fever present 48 hours after hospitalization, and high-grade obstruction on computed tomography (CT) scan were significant predictors of the need for surgery. Multivariable analysis revealed that persistent abdominal pain or distention (hazard ratio [HR], 3.04; P = 0.013), both persistent abdominal pain and distention (HR, 4.96; P < 0.001), fever at 48 hours (HR, 3.66; P = 0.038), and CT-determined high-grade obstruction (HR, 3.45; P = 0.017) independently predicted the need for surgery. Eighty-five per cent of patients with none of these four significant risk factors were successfully managed nonoperatively. Conversely, 92 per cent of patients with three or more risk factors required laparotomy. This analysis revealed four readily evaluable clinical parameters that may be used to predict the need for surgery in patients presenting with SBO: persistent abdominal pain, abdominal distention, fever at 48 hours, and CT findings of high-grade obstruction. These factors were combined into a predictive model that may of use in predicting failure of nonoperative SBO management. Early operation in these patients should decrease length of stay and diagnostic costs.