Objective NETL is a separate clinical subgroup of primary lung tumors and owing to the increased lung cancer screening,the prevalence of NETL has the fast growth.As a rare tumor,current TNM staging system has a limited role in prognostic stratification and treatment guidance for neuroendocrine tumor of the lung(NETL).The study aimed to explore a specific staging system for these tumors.Methods We retrospectively analyzed surgically-treated patients(T1a-4N0-2M0)who were diagnosed with NETL(includes typical carcinoid; atypical carcinoid large cell neuroendocrine carcinoma)between 2004 and 2015 from Surveillance,Epidemiologic,and End Results(SEER)database.Patients diagnosed before 2010 were set as training group while the others who were diagnosed after(include)2010 as validation group.Multivariable Cox regression analysis was used to identify factors associated with overall survival,and a new staging system was rearranged and staging groups were proposed by survival tree that formed by the stratified hazard ratios related to the H1N0.The predictability of proposed and 8th edition stages were evaluated by Cox regression analysis and we compared two staging system by Harrells C-statistics,net reclassification improvement and integrated discrimination improvement.Results 6450 patients included in the study(training group: validation group,2892:3558).In the training set,in the univariate analysis of the factors influencing survival,age,race,gender,histology,therapy,stage,T and N stage were all significant,however,hazard ratios between stage IA1 to ⅡA were close,even not significant.In multivariate Cox analysis,it showed that histology and N stage were independently associated with survival,but T stage wasnt.Therefore,we rearranged histological subtypes(H1,typical carcinoid; H2,atypical carcinoid H3,large cell neuroendocrine carcinoma)and N stage(N0,N1,N2)as follow: IA(H1N0); IB(H1N1,H2N0); Ⅱ(H1N2,H2N1-2,H3N0); Ⅲ(H3N1-2).Verified in the validation group,respective 5-year survivals were 91.6%,81.6%,53.7%,and 27.8%for IA to Ⅲ,however,the survival rates were 85.2%,83.9%,86.1%,78.9%,80.8%,70.3%,59.8%,71.0%for IA1-ⅢB based on 8th edition.The proposed staging system had a significant better predictability than 8th edition TNM staging system(C index: 0.750 [95%CI,0.736 to 0.764] vs0.617 [95%CI,0.600 to 0.634]),and,compared to the 8th edition,enhanced the identification of high-risk patients with a NRI 0.617(95%CI,0.561 to 0.672),better predicted differences in survival with a relative IDI 20.1%(95%CI,17.5%to 22.6%).Furthermore,in multivariate analyses,surgery+chemotherapy was a protective factor for survival,but in 8th edition TNM staging system,surgery+chemotherapy didnt confer a significantly better survival in any stage.Fortunately,in the new stage,adjuvant chemotherapy conferred a significantly better survival for stage Ⅱ and Ⅲ NELT after resection(p=0.043,p<0.001),but in proposed stage IA and IB,chemotherapy didnt provide survival benefit.Conclusions The new staging system can predict the prognosis of NETL better than 8th edition TNM stages and have the potential to guide postoperative treatments.However,the new staging system still needs external validation and more studies need to be done for optimizing the NETL-specific staging system.