Pancreatic endocrine tumors are rare diseases and devising a clinically effective prognostic stratification of patients is a major clinical challenge. This study aimed at assessing whether the tumor-node-metastasis (TNM)-based staging and proliferative activity-based grading recently proposed by the European NeuroEndocrine Tumors Society (ENETS) have clinical value. TNM was applied to 274 patients with histologically diagnosed pancreatic endocrine tumors operated from 1991 to 2005, with last follow-up at December 2007. According to World Health Organization (WHO) classification, 246 were well-differentiated neoplasms (51 benign, 56 uncertain behavior, 139 carcinomas) and 28 poorly differentiated carcinomas. Grading was based on Ki67 immunohistochemistry. Survival analysis not only ascertained the prognostic value of the TNM system but also highlighted that in the absence of nodal and distant metastasis, infiltration and tumor dimensions over 4 cm had prognostic significance. T parameters were then appropriately modified to reflect this weakness. The 5-year survival for modified TNM stages I, II, III and IV were 100, 93, 65 and 35%, respectively. Multivariate analysis identified TNM stages as independent predictors of death, in which stages II, III and IV showed a risk of death of 7, 29 and 58 times higher than stage I tumors (P<0.0001). Ki67-based grading resulted an independent predictor of survival with cut-offs at 5 and 20%. In conclusion, WHO classification assigns clinically significant diagnostic categories to pancreatic endocrine tumors that need prognostic stratification by applying a staging system. The ENETS-TNM provides the best option, but it requires some modifications to be fully functional. The modified TNM described in this study ameliorates the clinical applicability and prediction of outcome of the ENETS-TNM; it (i) assigns a risk of death proportional to the stage at the time of diagnosis, and (ii) allows a clinically based staging of patients, as the T parameters as modified permit their clinical-radiological recognition. Ki67-based grading discerns prognosis of patients with same stage diseases.