Background and Aims: Endoscopic ultrasound-guided fine needle aspiration (EUS-FNA) is a safe and accurate technique for diagnosing pancreatic cancer. The value of repeat EUS-FNA in patients with high clinical suspicion for pancreatic cancer after an inconclusive index study is unknown. Our aims were to determine the yield and success of repeat EUS-FNA and the reasons for failure of initial EUS-FNA. Methods: This was a retrospective analysis of prospectively collected data in a tertiary University based referral center for pancreatico-biliary disorders. All patients who underwent more then one EUS-FNA for evaluation of suspected pancreatic cancer over a five and a half year period were included in this analysis. Results: Of the 547 procedures performed on 517 patients, 24 (4.6%) patients underwent 51 repeat EUS-FNA procedures. Initial EUS-FNA was atypical/suspicious in 10 (41.6%), benign in 10 (41.6%), malignant in two (8.3%), and failed/indeterminate in two (8.3%) patients. Eight of 10 (80%) patients with atypical/suspicious findings at initial EUS-FNA were diagnosed with malignancy on repeat EUS-FNA. Of the 10 patients with benign findings at initial EUS-FNA, repeat study diagnosed two (20%) with malignancy and the rest were confirmed benign on long-term follow up (average 530 days, SD 369 days). Of the two patients with indeterminate findings at initial EUS-FNA, repeat study diagnosed one patient with malignant disease and the other with benign disease that was confirmed by long-term follow up. Of the two patients diagnosed with neoplastic disease at initial EUS-FNA, repeat EUS-FNA with immunostains downgraded both to chronic pancreatitis. Repeat EUS-FNA facilitated determination of the true status of disease in 20 of 24 patients (accuracy 84%). Suspected reasons for failed initial EUS-FNA were: coexisting pancreatitis (n = 10; 42%), technical difficulty due to scope positioning in uncinate lesion/sedation failure (n = 4; 16.7%), difficult cytology (partly cystic, extensive necrosis, well-differentiated adenocarcinoma) (n = 4; 16.7%), presence of ascites or collaterals (n = 3; 12.5%), pathologist's interobserver variation (n = 2; 8.33%), and unknown reason in one patient. Conclusion: Repeat EUS-FNA is warranted in patients with high clinical suspicion for pancreatic cancer despite indeterminate or negative findings at initial EUS-FNA. © 2007 The Authors.