BACKGROUND & AIMS: Endoscopic ultrasound (EUS)-guided fine needles with side fenestrations are used to collect aspirates for cytology analysis and biopsy samples for histologic analysis. We conducted a large, multicenter study to compare the accuracy of diagnosis via specimens collected with fine-needle biopsy (FNB) versus fine-needle aspiration (FNA) for patients with pancreatic and nonpancreatic masses. METHODS: We performed a prospective single-blind study at 5 tertiary care centers in China. The study comprised 408 patients undergoing EUS for a solid mass (>1 cm) in the pancreas, abdomen, mediastinum, or pelvic cavity, from December 2014 through January 2016. Patients were randomly assigned to groups (1:1) for assessment by FNA (n = 190) or FNB (n = 187). After lesions were identified by EUS, samples were collected in a total of 4 passes by each needle. All procedures were performed by experienced endosonographers; cytologists and pathologists were blinded to the sample collection method. Patients were followed for at least 48 weeks, and final diagnoses were obtained after surgery, imaging analysis, or resolution of lesion. The primary aim was to compare diagnostic yields of EUS-FNA with EUS-FNB for all solid masses, then separately as pancreatic and nonpancreatic masses. The secondary endpoint was the quality of histologic specimen. RESULTS: Findings from FNB analysis were accurate for 91.44% of all cases, compared with 80.00% for all FNA cases, based on final patient diagnoses (P = .0015). In patients with pancreatic masses (n = 249), findings from histologic analysis of FNBs were accurate for 92.68% of the cases, compared with 81.75% for FNAs (P = .0099). In cytology analysis of pancreatic masses, samples collected by FNB accurately identified 88.62% of all pancreatic lesions, whereas samples collected by FNA accurately identified 79.37% (P = .00468). Analyses of samples of nonpancreatic masses collected by FNA versus FNB produced similar diagnostic yields. CONCLUSIONS: In a prospective study of patients with pancreatic masses, we found EUS-guided FNB samples to produce more accurate diagnoses than samples collected by EUS-guided FNA samples. No difference in diagnostic yield was seen between EUS-FNA and EUS-FNB for nonpancreatic masses.