Predicting severe acute pancreatitis (AP) remains a challenge. The present study compares admission blood urea nitrogen (BUN), hematocrit, and creatinine, as well as changes in their levels over 24 h, aiming to determine the most accurate laboratory test for predicting persistent organ failure and pancreatic necrosis. Clinical data of 1,612 AP patients, enrolled prospectively in three independent cohorts (University of Pittsburgh, Brigham and Women's Hospital, Dutch Pancreatitis Study Group), were abstracted. The predictive accuracy of the studied laboratories was measured using area under the receiver-operating characteristic curve (AUC) analysis. A pooled analysis was conducted to determine their impact on the risk for persistent organ failure and pancreatic necrosis. Finally, a classification tree was developed on the basis of the most accurate laboratory parameters. Admission hematocrit ≥44% and rise in BUN at 24 h were the most accurate in predicting persistent organ failure (AUC: 0.67 and 0.71, respectively) and pancreatic necrosis (0.66 and 0.67, respectively), outperforming the other laboratory parameters and the acute physiology and chronic health evaluation-II score. In a pooled analysis, admission hematocrit ≥44% and rise in BUN at 24 h were associated with an odds ratio of 3.54 and 5.84 for persistent organ failure, and 3.11 and 4.07, respectively, for pancreatic necrosis. In addition, the classification tree illustrated that when both admission hematocrit was ≥44% and BUN levels increased at 24 h, the rates of persistent organ failure and pancreatic necrosis reached 53.6% and 60.3%, respectively. Admission hematocrit ≥44% and rise in BUN at 24 h may be the optimal predictive tools in clinical practice among existing laboratory parameters and scoring systems.