Reporting of Ki67 labeling index (LI) is a routine in diagnostics of neuroendocrine neoplasms of the pancreas. The aim of the study was to examine whether heterogeneity of Ki67 LI distribution in primary tumoral tissue influences precision of reporting of Ki67 LI and Ki67-LI-based grade, both established in adherence to WHO 2010 guidelines. Seventy-one samples of neuroendocrine tumours (NET) and 6 samples of neuroendocrine carcinomas (NEC) of the pancreas were taken for manual counting of Ki67 LI in 25 portions of 100 cells (2500 cells in total) in 3 hot spots an in a single area of lower proliferation rate (cold spot) in each case. Both NET and NEC showed Ki67 LI heterogeneity within primary tumour. Almost 20% of NET showed higher grade when 500 cells rather than 2000 cells were counted in hot spot area. Suboptimal choice of hot spot resulted in under-grading of approximately 20% of NET. Cold spots were constantly present in NET. Heterogeneity of Ki67 LI was also present in NEC, but it virtually never resulted in under-grading. Concept and methodology of Ki67 LI counting in neuroendocrine neoplasms of the pancreas requires clarification. Efforts aiming to improve precision of assessment of Ki67 LI are needed.