Purpose: Endoscopic ultrasound-guided fine-needle aspiration (EUS-FNA) is an important tool for assessing pancreatic masses. CT-guided core biopsy (CT-CB) is also widely used for this purpose. We examined the accuracy, risk and costs of these two modalities. Methods: We reviewed all patients who underwent either EUS-FNA or CT-CB for pancreatic masses seen on CT scan at one institution (VMMC) between January 2005 and September 2008. Results were categorized as: negative, atypical, positive or suspicious for malignancy. Final diagnosis was confirmed by histologic examination of surgical specimens or clinical follow-up. Diagnostic yields were compared and costs were calculated using 2008 Medicare rates. Results: A total of 344 patients underwent EUS-FNA and 66 patients underwent CT-CB. 20 EUS-FNA patients and 4 CT-CB patients were excluded due to inadequate follow-up or uncertain final diagnosis. In the 260 EUS-FNA patients ultimately diagnosed with malignancy, histology was positive in 219 (84%), suspicious for malignancy in 15 (6%), and negative or atypical in 26 cases (10%). Malignancies included adenocarcinoma (86%), neuroendocrine tumors (8%), lymphoma (3%), and metastases from kidney, lung, and breast (2%). Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and diagnostic accuracy were 84%, 100%, 100%, 51%, and 86%, respectively. All patients with suspicious cytology were ultimately proven to have cancer. Inclusion of these cases increased sensitivity to 90%. Among the 47 CT-CB patients found to have cancer, histology was positive for malignancy in 35 cases (sensitivity 74%) with no false positive results (specificity 100%). PPV, NPV, and diagnostic accuracy were 100%, 56%, and 84%, respectively. However, 22 of these patients who underwent CT-CB had undergone prior negative EUS-FNA. When these cases are excluded, CT-CB sensitivity increases to 93%. There was a trend toward a difference in sensitivity between the EUS and CT groups (p=0.06) when these 22 CT-CB patients are considered and when they are excluded (p=0.07). There was one bleed in the CT-CB group (1.6%) and 3 cases of pancreatitis in the EUS-FNA group (0.8%) (p=0.54). There was no procedure related mortality in either group. Using 2008 Medicare payment data and assuming a negative EUS-FNA or CT-CB case could be resolved with a second procedure using the other modality, the cost of a positive diagnosis using EUS-FNA was $1,078 versus $1,155 for CT-CB. Conclusion: Both EUS-FNA and CT-CB are safe and accurate methods for tissue diagnosis in patients with solid pancreatic lesions. EUS-FNA appears to be at least as accurate as CT-CB and less expensive.