Main pancreatic duct dilatation raises concerns about the possibility of pancreatobiliary malignancy. We evaluated the etiologic yield of endosonography (EUS) for main pancreatic duct dilatation without definite pathology on Ultrasonography (US).A retrospective review was conducted in 54 consecutive patients referred for EUS.No pathological finding (37.0%, 20/54), followed by periampullary cancer (35.2%, 19/54), was the most common finding. Elevated alkaline phosphatase (ALK-P) and marked common bile duct (CBD) dilatation (≥ 12 mm) were the predictors of malignancy (p < 0.05). Among the 37 subjects with available ALK-P and CBD diameter, the probability of malignancy was 84.6% (11/13) for both elevated ALK-P and marked CBD dilatation, 16.7% (1/6) for isolated elevated ALK-P, 18.2% (2/11) for isolated marked CBD dilatation, and none (0/7) was for subjects with neither elevated ALK-P nor marked CBD dilatation, respectively. The overall accuracy of EUS for periampullary carcinomas was 94.7% (18/19) and for choledocholithiasis was 100% (7/7), respectively. EUS had a 100.0% (20/20) sensitivity and a 97.1% (33/34) specificity in the diagnosis of no pathological obstruction.EUS is accurate for main pancreatic duct dilatation without definite pathology on US, and the presence of concomitant elevated ALK-P and CBD dilatation highly suggests malignancy.