Background: The objective of our study was to assess a single operator's learning curve with regard to the number of passes, the diagnostic accuracy, and the complications associated with EUS-guided FNA (EUS-FNA) of solid pancreatic masses. Methods: The number of passes, the diagnostic accuracy, and the complication rate were prospectively evaluated in 300 consecutive EUS-FNA of solid pancreatic masses performed by a single endosonographer over a 3-year period. The procedures were placed into 3 groups, which contained 100 procedures each. The endosonographer had undergone a third-tier EUS fellowship and had performed 45 supervised pancreatic EUS-FNA during his training. Results: Of the 300 EUS-FNA performed (median age 63 years, 64% men), no statistically significant differences among the 3 groups with regard to age, gender, race, location, or size of the mass were found. Diagnostic accuracy of the EUS-FNA procedure was similar over time (Group 1, 92%; Group 2, 92%; Group 3, 95%). Median number of passes showed a decreasing trend over the 3-year study period, despite an increasing trend of the number of procedures performed (r = -0.14, p = 0.42). The median number of passes was lower for Group 2 (median, 3; p = 0.02) and Group 3 (median, 3; p = 0.003) compared with Group 1 (median, 4). Group 3 (7/100, 7%) was less likely to encounter complications compared with Group 1 (13/100, 13%; p = 0.24) and Group 2 (18/100, 18%; p = 0.03). Frequency of serious complications was similar across the 3 groups (1%-3%). Conclusions: With adequate third-tier training, a newly developed EUS program can achieve safe and accurate results of EUS-FNA of the pancreas. The learning curve, however, needs to continue after the fellowship, because more procedures are needed for one to gain proficiency and efficiency with EUS-FNA. Copyright © 2005 by the American Society for Gastrointestinal Endoscopy.