Background The diagnostic yield of endoscopic ultrasound (EUS) guided fine-needle aspiration (FNA) is variable, and partly dependent upon rapid onsite evaluation (ROSE) by a cytopathologist. Second generation fine-needle biopsy (FNB) needles are being increasingly used to obtain core histological tissue samples. Aims Studies comparing the diagnostic yield of EUS guided FNA versus FNB have reached conflicting conclusions. We therefore conducted a systematic review and meta-analysis to compare the diagnostic yield of FNA with FNB, and specifically evaluating the diagnostic value of ROSE while comparing the two types of needles. Methods We searched several databases from inception to 10 April 2016 to identify studies comparing diagnostic yield of second generation FNB needles with standard FNA needles. Risk ratios (RR) were calculated for categorical outcomes of interest (diagnostic adequacy, diagnostic accuracy, and optimal quality histological cores obtained). Standard mean difference (SMD) was calculated for continuous variables (number of passes required for diagnosis). These were pooled using random effects model of meta-analysis to account for heterogeneity. Meta-regression was conducted to evaluate the effect of ROSE on various outcomes of interest. Results Fifteen studies with a total of 1024 patients were included in the analysis. We found no significant difference in diagnostic adequacy [RR 0.98 (0.91, 1.06), (I2 = 51 %)]. Although not statistically significant (P = 0.06), by meta-regression, in the absence of ROSE, FNB showed a relatively better diagnostic adequacy. For solid pancreatic lesions only, there was no difference in diagnostic adequacy [RR 0.96 (0.86, 1.09), (I2 = 66 %)]. By meta-regression, in the absence of ROSE, FNB was associated with better diagnostic adequacy (P = 0.02). There was no difference in diagnostic accuracy [RR 0.99 (0.95, 1.03), (I2 = 27 %)] or optimal quality core histological sample procurement [RR 0.97 (0.89, 1.05), (I2 = 9.6 %)]. However, FNB established diagnosis with fewer passes [SMD 0.93 (0.45, 1.42), (I2 = 84 %)]. The absence of ROSE was associated with a higher SMD, i. e., in the presence of an onsite pathologist, FNA required relatively fewer passes to establish the diagnosis than in the absence of an onsite pathologist. Conclusions There is no significant difference in the diagnostic yield between FNA and FNB, when FNA is accompanied by ROSE. However, in the absence of ROSE, FNB is associated with a relatively better diagnostic adequacy in solid pancreatic lesions. Also, FNB requires fewer passes to establish the diagnosis.