The aim of this study is to elucidate in which subset of patients chemotherapy offers a real benefit. Methods: We retrospectively investigated the US National Cancer Data Base and evaluated patients who underwent pancreatic resection (pancreaticoduodenectomy, partial and distal pancreatic resection) for PNET from 2004 to 2014. Differences between two groups, Chemotherapy Group (CH), and Surgery Alone Group (S) were analyzed, and comparison was performed to identify the beneficial effect of chemotherapy. Study end point was overall survival. Results: Population study included 13822 patients, 8259 were in Group CH and 5563 were in Group S The two groups were well matched in demographic profile including age, race distribution, co-morbidities (Charlson/Deyo Score), pathological stage, grading and status of resection margins. Multivariable Cox proportional hazard regression model demonstrated pathological grade (HR1.680), positive surgical margin (HR 1.403), local invasion (HR 1.645) associated with overall survival, instead lymphovascular invasion (HR 1.030, NS) and positive node (HR 1.175) have a marginal effect on it. Kaplan-Meier analysis demonstrated no superior survival for Chemotherapy group when compared to Surgery alone group despite positive node (mean OS 27.3 months vs 27.8 months, NS),positive surgical margin (mean OS 27.3 months vs 28.5 months, NS), positive lymphovascular invasion (mean OS 28.4 months vs 36.1 months, NS) and local invasion (mean OS 29.0 months vs 29.9 months, NS). Chemotherapy has a limited effect only with poorly/undifferentiated lesion (mean OS 25.5 months vs 17.6 months, p < .005). Conclusion:Current chemotherapy regimen has a beneficial effect limited to high grade lesion. Extensive surgical resection is the only treatment that offers a survival benefit even for cancer that presents with aggressive features including positive margins, local invasion, positive lymphnode and presence of lymphovascular invasion.