BACKGROUND AND STUDY AIMS Endoscopic retrograde cholangiopancreatography (ERCP) has a significant mortality, morbidity, and failed cannulation rate. Magnetic resonance cholangiopancreatography (MRCP) is a safer, noninvasive method of imaging the pancreaticobiliary tree. A substantial number of patients are referred for ERCP because of abdominal pain, a high proportion of whom have normal ducts or pathology not requiring interventional ERCP. The aim was to assess the potential impact of MRCP on overall ERCP workload and patient outcome if MRCP were the primary investigation in patients referred for ERCP because of abdominal pain. PATIENTS AND METHODS 1758 consecutive ERCPs performed in 1148 patients over a 3-year period in a single tertiary referral center in the pre-MRCP era were reviewed. Cannulation failure, ERCP findings, need for follow-up ERCP and all 30-day major complication rates were analyzed with regard to clinical indications. RESULTS The overall workload comprised 1108 (63 %) successful initial ERCPs, 188 (11 %) failed cannulation attempts and 462 (26 %) follow-up ERCPs. Of the patients, 299 (27 %) had normal ERCP findings, 331 (30 %) had choledocholithiasis and 246 (22 %) had strictures. lf MRCP had been used as the primary imaging investigation in the 451 patients (39 %) referred for ERCP because of abdominal pain, we estimate that 197 patients (44 %) would have avoided ERCP, and the overall ERCP workload would have been reduced by 13 %. Initial MRCP in suspected gallstone pancreatitis and certain miscellaneous groups, it was estimated, would have further decreased ERCP workload by 9 %. Four of 40 major ERCP-related complications (3.5 %) and one of four ERCP-related deaths (0.35 %) would potentially have been avoided. CONCLUSIONS Initial MRCP in patients referred with abdominal pain would potentially have avoided ERCP in 44 % of cases, reduced ERCP workload by 13 % and significantly reduced patient morbidity and mortality. The relatively small reduction in ERCP workload among these patients reflects the fact that over half of them had probable sphincter dysfunction, a significant proportion of whom might have benefited from biliary manometry and/or endoscopic intervention despite a normal MRCP. Furthermore, a small number of patients with calculi and subtle biliary and pancreatic strictures would be missed by this approach.