Background and study aims Pancreatic neuroendocrine tumors (PanNETs) outcomes are dependent upon grading by Ki67. This study compared endoscopic ultrasound-guided fine-needle biopsy (EUS-FNB) to fine-needle aspiration (FNA) in assessing PanNETs. Patients and methods All pancreatic histology for PanNET between January 2009 and June 2017 was included if EUS sampling was performed prior to surgical resection. Ki67 and grade from FNA and FNB samples was compared to surgical histology using correlation coefficient and kappa values. Subgroup analysis was performed for purely solid lesions, lesions < 2 cm and FNB needle type. Results One hundred sixity-four patients had PanNET of which 57 underwent surgical resection. Thirty-five lesions underwent FNA and 26 FNB (4 had both) confirming PanNET. 23/ of 35 FNA samples reported Ki67/grading compared to all 26 FNB samples ( P = 0.0006). Compared to surgical histology, Ki67 on FNA correlated poorly overall (r = -0.08), in solid lesions (r = -0.102) and lesions < 2 cm (r = -0.149) whereas FNB correlated moderately overall (r = 0.65), in solid lesions (r = 0.64) and lesions < 2 cm (r = 0.61). Tumor grade showed poor agreement (kappa) with FNA overall (0.026), in solid lesions (0.044) and lesions < 2 cm (0.00) whereas FNB showed moderate-good agreement overall (0.474), in solid lesions (0.58) and lesions < 2 cm (0.745). Fork-tip FNB needles Ki67 showed strong correlation with surgical histology (r = 0.788) compared to reverse bevel FNB needles (r = 0.521). Both FNB needles showed moderate agreement with tumor grade. Conclusion FNB samples were significantly more likely than FNA to provide adequate material for Ki67/grading and showed a closer match to surgical histology. FNB needle types require prospective investigation.