BACKGROUND: Endoscopic ultrasound-guided fine-needle aspiration (EUS-FNA) has emerged as the diagnostic modality of choice for mass lesions in the pancreas. The objective of the current study was to determine the accuracy and pitfalls of EUS-FNA in the diagnosis of pancreatic lesions in cases that involved follow-up surgical resection. METHODS: Cases of EUS-FNA of pancreatic lesions performed from 2007 to mid-2012 for which subsequent surgical resection was performed were retrieved from the department's database. The accuracy of the cytologic diagnosis was assessed using the histological diagnosis as the gold standard. The sensitivity, specificity, positive predictive value, and negative predictive value were calculated. "Neoplastic," "suspicious," and "malignant" were classified as a positive cytologic diagnosis. In one calculation method, "atypical" was also included as a positive cytologic diagnosis whereas in another it was not considered to be a positive cytological result. The cases with a cytologic-histological discrepancy were reviewed to identify sources of errors.RESULTS: A total of 1212 cases from 1104 patients (518 women and 586 men; age range, 18-94 years [average age, 63.5 years]) were identified. Cytologic diagnoses included 52 unsatisfactory, 224 benign, 129 atypical, 140 neoplasm, 35 suspicious, and 632 malignant diagnoses. Of these cases, 397 patients had histological follow-up information available. The sensitivity, specificity, positive predictive value, and negative predictive value were 83.2%, 85.9%, 95.9%, and 56.1%, respectively, with atypical cases excluded from the analysis. When atypical cases were included as a positive cytologic diagnosis, the sensitivity, specificity, positive predictive value, and negative predictive value were 86.7%, 67.9%, 90.7%, and 58.5%, respectively, and were 73.7%, 87.7%, 95.6%, and 48.0%, respectively, when atypical cases were included as a negative cytologic diagnosis. The major difficulty in EUS-FNA cytology was to differentiate pancreatic mucinous neoplasms from contaminants of gastric mucosa. Other pitfalls included differentiating mucinous neoplasm from extensive pancreatic intraepithelial neoplasia, and endocrine tumor from nesidioblastosis versus acinar cell carcinoma or intrapancreatic spleen.CONCLUSIONS: EUS-FNA is a valuable tool for the diagnosis of pancreatic lesions, especially solid malignant tumors.Cytologic-radiological correlation is essential in differentiating pancreatic mucinous neoplasms from gastric mucosa, because the former usually are found to have characteristic features on imaging. Pathologists should be aware of the pitfalls in the cytologic diagnosis of pancreatic lesions that may significantly change the clinical management of the patients.Cancer (Cancer Cytopathol) 2015;123:98-107.