BACKGROUND: The aim of this study was to describe the spectrum of disease in children with acute pancreatitis and assess predictors of severity. METHODS: Children (= 18 years) admitted to a single institution with acute pancreatitis from 2000 to 2009 were included. The accuracy of the Ranson, modified Glasgow, and pediatric acute pancreatitis severity (PAPS) scoring systems for predicting major complications was assessed. RESULTS: The etiology of pancreatitis in these 211 children was idiopathic (31.3%), medication-induced (19.9%), gallstones (11.8%), trauma (7.6%), transplantation (7.6%), structural (5.2%), and hemolytic-uremic syndrome (3.3%). Fifty-six patients (26.5%) developed severe complications. Using the cutoff thresholds in the PAPS scoring system, only admission white blood cell count more than 18,500/muL (odds ratio [OR], 3.1; P = .010), trough calcium less than 8.3 mg/dL (OR, 3.0; P = .019), and blood urea nitrogen rise greater than 5 mg/dL (OR, 4.1; P = .004) were independent predictors of severe outcome in a logistic regression model. The sensitivity (51.8%, 51.8%, 48.2%) and negative predictive value (83.2%, 83.5%, 80.5%) of the Ranson, modified Glasgow, and PAPS scores were, respectively, insufficient to guide clinical decision making. CONCLUSION: Commonly used scoring systems have limited ability to predict disease severity in children and adolescents with acute pancreatitis. Careful and repeated evaluations are essential in managing these patients who may develop major complications without early signs.