The APACHE II score is highly recommended worldwide for the assessment of severe pancreatitis (interstitial and necrotizing), and a score of at least eight points on admission to the hospital is said to indicate severe pancreatitis. To evaluate this assumption and to check whether an APACHE II score of at least eight points really indicates necrotizing pancreatitis as shown by contrast-enhanced computed tomography (CT). This study included 326 patients with a first attack of acute pancreatitis and is part of a prospective study on the natural course of acute pancreatitis. All patients underwent contrast-enhanced CT within 72 hours of admission. The following parameters for the severity of the disease were used: respiratory and renal failure according to the Atlanta classification; indication for dialysis, ventilation, and surgery; time spent in intensive care unit and total hospital stay; Ranson score adjusted for cause; Imrie score; and Balthazar score (CT). Of the 326 patients, 262 (80%) had interstitial pancreatitis and 64 (20%) had necrotizing pancreatitis. In 74 (28%) of the 262 patients with interstitial pancreatitis, the APACHE II score was at least eight points, indicating severe pancreatitis (overestimation of the disease), whereas the score was less than eight in 41 (64%) of 64 patients with necrotizing pancreatitis (underestimation). Sensitivity was 36%; specificity was 72%; the positive predictive value was 24%; and the negative predictive value was 82%. The evaluation of sensitivity, specificity, and positive and negative predictive value for all APACHE II score points showed that there was not a "golden" cutoff to detect necrotizing pancreatitis. We conclude that the APACHE II score on admission to the hospital is unreliable to diagnose necrotizing pancreatitis.