BACKGROUND: Needle aspiration of the pancreas is performed to differentiate pancreatic malignancy, focal chronic pancreatitis, and metastasis to the pancreas. Biopsies may be directed by using EUS, CT, US, or surgery. This study retrospectively compared the accuracy of EUS-guided, CT/US-guided, and surgical tissue sampling of the pancreas over a 5-year period. METHODS: The records of patients undergoing pancreatic tissue sampling were reviewed for a final clinical diagnosis based on the results of cytology, histology, and clinical history. The sensitivity, specificity, and accuracy of each technique were calculated. RESULTS: One hundred forty-nine tissue samples (68 EUS-guided, 70 CT/US-guided, 11 surgical) from 128 patients were compared. There was no significant difference in accuracy rates for EUS (76.4%), CT/US (81.4%), and surgically guided (81.8%) specimens. EUS was used when masses were smaller (2.6 +/- 0.1 cm) as compared with CT/US (3.4 +/- 0.2 cm, p < 0.001) and surgery (2.9 +/- 0.4 cm, p = 0.49). In univariate analyses, factors associated with greater accuracy regardless of technique were as follows: (1) older age, (2) larger size of the mass, and (3) participation by a cytologist during the procedure. A subsequent multivariate logistic regression analysis, in which the examination of the effect of each factor controls for the effect of each of the other factors, found that only older age was a significant predictor of accuracy. CONCLUSION: EUS-guided tissue sampling of pancreatic masses is as accurate as CT/US-guided sampling and surgical biopsies.