156 Background: Adopting a unified staging system for pancreatic neuroendocrine tumors (PNET) has been challenging. Currently, the American Joint Committee on Cancer (AJCC) recommends the use of the pancreatic adenocarcinoma staging system for PNET. We sought to validate this recommendation on a large administrative population database. Methods: Surveillance, Epidemiology, and End Results (SEER) data were used to identify patients with PNET (excluding patients with large cell, small cell, or mixed endocrine-exocrine carcinoma) who underwent curative-intent surgical resection from 1983 to 2008. The discriminatory ability of the AJCC system (recorded by SEER since 2004) was examined and a new TNM system was devised utilizing extent of disease variables. Results: Of 1,202 patients identified, 51% were female. Median age was 55 years (range, 9-93). Lymph node metastasis (present in 43% of patients) was associated with worse overall survival (OS) after resection (10-year OS, 50% vs 63%, p<.0001), as was the presence of distant metastasis (present in 24% of patients, 10-year OS, 35% vs 63%, p<.0001).The current AJCC system (available in 412 patients) distinguished overall survival adequately only between stages I and II, but not between II and III, or III and IV (Table). By modifying the T stage to be based only on size (0-1 cm, 1-2 cm, 2-4 cm, and > 4 cm) and by revising the grouping allocation, we propose a novel TNM system with improved discriminatory ability (Table). Conclusions: In this study validating the current AJCC staging system for PNET, we found stages II, III, and IV to perform similarly. We propose a simplified TNM system that better discriminates between outcomes. [Table: see text]