Background and study aims: It is uncertain if needle gauge impacts the diagnostic accuracy of endoscopic ultrasound-guided fine needle aspiration (EUS-FNA) of pancreatic mass lesions. Our aim was to use meta-analysis to more robustly define the diagnostic accuracy of EUS-FNA for pancreatic masses using 22 G and 25 G needles. Patients and methods: Studies were identified by searching nine medical databases for reports published between 1994 and 2011, using a reproducible search strategy comprised of relevant terms. Only studies comparing the overall diagnostic accuracy of 22 G vs. 25 G EUS needles that used surgical histology or at least 6 months clinical follow up for a gold standard were included. Two reviewers independently scored the identified studies for methodology and abstracted pertinent data. When required, the original investigators were contacted to provide additional data. Pooling was conducted by both fixed-effects and random-effects models. Diagnostic characteristics (sensitivity, specificity, positive and negative likelihood ratios) with 95 % confidence intervals (CIs) were calculated. Results: Eight studies involving 1292 subjects met the defined inclusion criteria. Of the 1292 patients, 799 were in the 22 G group and 565 were in the 25 G group (both needles were used in 72 patients). The pooled sensitivity and specificity of the 22 G needle were 0.85 (95 %CI 0.82 - 0.88) and 1 (95 %CI 0.98 - 1) respectively. The pooled sensitivity and specificity of the 25 G needle were 0.93 (95 %CI 0.91 - 0.96) and 0.97 (95 %CI 0.93 - 0.99) respectively. The bivariate generalized linear random-effect model indicated that the 25 G needle is associated with a higher sensitivity (P = 0.0003) but comparable specificity (P = 0.97) to the 22 G needle. Conclusions: This meta-analysis suggests 25 G needle systems are more sensitive than 22 G needles for diagnosing pancreatic malignancy.