Background: NETs often present as liver metastasis with an unknown primary. Accurate subtyping of NETs has important clinical implications for staging and site-specific targeted therapy. Traditionally, the work-up to identify a primary NET as being lung, pancreatic or gut-based can be challenging and time-consuming. Methods: GEP was performed on formalin-fixed, paraffin-embedded tumor samples using a 92-gene RT-PCR assay (CancerTYPE ID, bioTheranostics Inc.) as part of the clinical work-up for patients diagnosed with NETs and unknown primaries. Results: Results were categorized by level of agreement (Table). Of the 39 patients tested with the assay, 82% presented with liver metastasis. Assay results from those patients with adequate work-up were concordant with clinical data in 77% (23/30) of cases. Surgery was performed in 12 of these cases and 100% accuracy of the molecular assay was confirmed, resulting in 75% of primary tumors being found in the gut and 25% in the pancreas or duodenum. Assay predictions were clinically plausible but inconsistent in 13% (4/30) of cases and were discordant with histology, IHC, imaging/radiological findings, and clinical impression in 10% of the cases. Conclusions: The 92-gene assay accurately predicted tumor subtype in patients presenting with NETs and an unknown primary. These findings have clinical utility for appropriate treatment selection, particularly where targeted therapies are available (everolimus, sunitinib). We believe the 92-gene assay can be useful in clinical management, and that our approach will lead to effective diagnosis and treatment algorithms to streamline extensive pre-operative work-up.