BACKGROUND/AIMS: Severe acute pancreatitis occurs in about 20% of the patients with acute pancreatitis and can be associated with multiorgan failure and local complications. In patients with predicted severe acute pancreatitis, overall mortality rates are about 15-30%. The aim of this study was to determine the factors correlated with mortality in patients with severe acute pancreatitis. METHODS: We reviewed five hundread and seventy two consecutive cases of acute pancreatitis from January, 2000 to December, 2005. Of them, 109 patients who fulfilled the criteria of Atlanta classification for severe acute pancreatitis were enrolled. Data were collected by chart reviews including age, gender, etiology, body mass index (BMI), modified Glasgow score, APACHE II score, APACHE III score, Balthazar CT index, and other laboratory parameters performed within 48 hours after the initial admission. RESULTS: Severe acute pancreatitis was most commonly caused by alcohol. Overall mortality rate was 20.2% in severe acute pancreatitis and 10 (45%) deaths occurred within the first week. Multiple logistic regression analysis identified serum creatinine, corrected calcium concentrations, and CT index as predictors of mortality in patients with severe acute pancreatitis. The risk score (R) was calculated by combining 3 prognostic values with regression coefficients; R=2.512 log(e) (creatinine mg/dL)+1.729 log(e) (CT index)-4.780 log(e) (corrected calcium mg/dL). The AUC for this score was 0.877 and a cutoff level of 0 was determined to predict the mortality with 83.3% sensitivity and 89.5% specificity. CONCLUSIONS: The newly designed risk score comprising 3 parameters can be used as the significant early predictor for hospital mortality in severe acute pancreatitis.