To clarify the curved linear array endoscopic ultrasonography (CLAEUS) findings that would enable prediction of cases with difficult selective biliary cannulation during endoscopic retrograde cholangio-pancreatography (ERCP). In this single-institution retrospective observational study, patients who underwent CLAEUS prior to an initial ERCP were examined. Patients who fulfilled the definition of difficult selective biliary cannulation were placed into a refractory (R) group and patients where biliary cannulation was completed within 20 minutes were placed into a straightforward (S) group. Difficult selective biliary cannulation was defined as when more than 20 minutes were required from the time to view the papilla to successful biliary cannulation, when the bile duct could not be cannulated, when the precut technique was necessary, when a CLAEUS-guided intervention was necessary, or when a transhepatic approach was necessary. We retrospectively compared CLAEUS findings, side-viewing scope findings, and other factors between the R and S groups as following. Variables examined: patient characteristics: 1) age, 2) gender, 3) diagnosis, and 4) gallbladder presence or absence; papillary findings on a side-viewing scope: 1) papillary type, 2) papillary size, and 3) periampullary diverticulum presence or absence; CLAEUS findings: 1) bile duct collapse with probe on regular evaluation at 2nd portion of duodenum (D2), 2) coaxial depiction of the cholangio-pancreatic duct at D2, and 3) background pancreas. A p value of <0.05 was considered statistically significant. The subjects were selected from 986 patients who underwent ERCP in the 36 months from July 1, 2014, to June 30, 2017. From these, 98 patients were selected who underwent naïve ERCP after CLAEUS. Of these, 17 and 81 were placed in the R (10 men, mean age 73.0±13.2 years, median 74 years) and S (42 men, mean age 67.6±15.1 years, median 70 years) groups, respectively (p=0.15). A significant difference was observed for bile duct collapse (p=0.02) and coaxial depiction of the cholangio-pancreatic duct (p<0.0001) on univariate analysis. The sensitivity, specificity, positive predictive value, and negative predictive value were 0.71, 0.87, 0.54, and 0.93, respectively. A significant difference was observed between ROC curves to predict difficult biliary cannulation by using bile duct collapse (AUC=0.55), and the factor plus coaxial depiction of the cholangio-pancreatic duct (AUC=0.81; p<0.0001). A finding of coaxial depiction of the cholangio-pancreatic duct on CLAEUS can be used to predict difficult ERCP cannulation.