Introduction: CT is considered the standard imaging modality in evaluation of patients with pancreatic head tumors; MRI technology is under evaluated and has some potential advantages for these patients.Methods: We reviewed all consecutive MRI exams obtained between 05/2004 and 10/2008 for patients with suspicion of adenocarcinoma of the head of the pancreas. All clinical decisions were based on MRI findings, and all patients underwent Whipple procedures.Results: 129 patients underwent MRI during the study period. 37 were considered unresectable (mets or locally advanced) and were not offered surgical treatment. 52 patients were considered resectable (R group)(minimal invasion of PV/SMV or not at all) and 40 were borderline resectable (B group) (abuttment -not encroachment- of PV/SMV or minimal arterial involvement). From R group, 38 were resected (3 with associated vascular resection), 5 had positive margins (13%), 4 unsuspected metastases (8%), 6 vascular involvement precluding R0 resection, 21 with positive lymph nodes (68%). From B group, 26 were resected (6 with vascular resection), 8 had positive margins (32%), 1 unsuspected metastases (3%), 7 vacular involvement precluding R0 resection, 17 with positive lymph nodes (55%). 2 patients had final diagnosis of chronic pancreatitis (1 resected, 1 unresected, both from B group). There was no difference (R vs B) on incidence of positive lymph nodes (p=0.45), unsuspected metastases (p=0.40) and a trend for positive margins on B (p=0.11) (All Fisher's Exact Test). Only 6 patients from both groups (R+B) had unsuspected metastases (7%).Conclusion: MRI is accurate for predicting presence of pancreatic adenocarcinoma (head), and the absence of intrabdominal spread of the disease in this selected group of patients. Vascular involvement and positive margins are less accurately determined by MRI. MRI may be the only imaging study needed to offer surgical treatment to these patients.