Introduction The World Health Organization (WHO) has provided grading guidelines for pancreatic neuroendocrine neoplasms (PanNENs) based on mitotic count and Ki-67 proliferation index. Driven by the desire to provide earlier tumor grading for clinical management decisions, some groups have proposed grading PanNENs at the time of fine-needle aspiration (FNA) using Ki-67 proliferation rates. Although a Ki-67 rate can be performed on FNA cell blocks, there are potential sampling limitations with this technique that may affect the reliability of the Ki-67 result. Materials and Methods Forty-nine PanNENs with FNA cell blocks and corresponding resection material were evaluated by immunohistochemistry for expression of Ki-67. Ki-67 proliferation rate was calculated based on cell counts >500 cells in the highest-staining areas. Ki-67 scores from FNA cell blocks were correlated with Ki-67 scores from resection specimens. Results The FNA Ki-67 proliferation rates overall did not correlate well with the resection specimen. A linear regression analysis of the correlation between FNA %Ki-67 and resection %Ki-67 showed a slope of 3.2 and an R2 = 0.58. The average difference in Ki-67 proliferation rate between FNA and resection was 5.9%. Thirty-nine percent (19 of 49 cases) of PanNENs showed discordant grading between the FNA cell block and resection specimen. Almost all (18 of 19) discordant cases demonstrated a lower FNA-based grade than the resection grade. Conclusions FNA cell block grading using Ki-67 rates frequently led to under-grading of the tumor. This finding is consistent with concerns that FNA may not provide accurate grading because of the limited sampling of the tumor.