In a prospective evaluation of 58 consecutive patients referred for operation of a suspected pancreatic or peri-ampullary cancer, the accuracy of ultrafast magnetic resonance imaging (UMRI) in predicting the resectability of pancreatic tumors compared with alternative staging interventions was assessed. The staging methods included: 1) transcutaneous ultrasound (US) with color Doppler, 2) UMRI, including echoplanar sequences and breath-hold gadolinium-enhanced dual-phase three-dimensional magnetic resonance angiography (MRA), 3) rapid bolus dual-phase helical computed tomography (CT), 4) angiography of celiac and mesenteric arterial systems, including portal venous phase, and 5) endoscopic cholangiopancreatography (performed in jaundiced patients). Patients were evaluated for extrapancreatic tumor spread, presence of hepatic metastases, lymph node involvement, and vascular involvement--each a sign of unresectability. After an investigator blinded to the results of the other imaging studies assessed resectability, patients were then divided into three categories: 1) probably resectable, 2) probably unresectable, and 3) certainly inoperable. Final diagnosis was obtained by laparotomy (47 of 58 pts), or by histopathological examination of fine needle aspiration specimens in patients deemed inoperable. The 58 suspected tumors were localized to the pancreatic head in 35 (60%), body in 11 (19%), and tail in one (2%). Nine (16%) ampullary tumors and two (3%) distal common bile duct tumors made up the remainder. For those 52 patients for whom histology was obtained, 44 were malignant and eight benign. Accuracy for assessing extrapancreatic tumor extension was highest with UMRI (95.7%) followed by US (85.1%), and CT (74.4%). UMRI provided the best means for detecting liver metastases with an accuracy of 93.5% compared with 87.2% for each of US and CT. UMRI, US, and CT had a reduced capacity for detecting lymph node involvement (80.4%, 76.6%, and 69.2%, respectively). In assessing vascular invasion, UMRI had an accuracy of 89.1%, US 83.0%, CT 79.5%, and angiography 68.8%. The findings suggest that UMRI is equal to or superior to other staging methods with regards to sensitivity, specificity, and overall accuracy. Since UMRI has the potential to reduce patient time, money, and discomfort, this study concludes that this staging technique should replace alternative methods as it provides an "all-in-one" diagnostic modality.