Objective Classification of non-small cell lung lymph (NSCLC) node (N) is one of the key factors influencing treatment, however, the cilinical noninvasive and invasive approaches to N classification have their limitations.This study aimed to investigate the risk factors of lymph node metastasis of peripheral lungadenocarcinoma by using CT and PET / CT scans.Methods Retrospective analysis had been done on a total of 248 patients who underwent surgical resection from February 2010 to November 2015 in our hospital.All of them underwent chest CT and 80 patients underwent PET/CT examination.Univariate analysis was applied in the relation of lymph node metastasis to gender, age, smoking situation, CEA, SUV, cancer size, pathological variants, and the degree of differentiation.Multivariable logistic regression analysiss were performed in the prediction of risk factors for lymph node metastasis.ResultsSeventy-four patients (29.8%) had regional lymph node metastases.Univariate analysis showed that lymph node metastasis was related to the serum CEA level, degree of differentiation, SUVmax, tumor size, lobulation/spiculation, pleural retraction, mediastinal or hilar lymphadenopathy (P<0.05).In the multivariable analysis of risk factors, including serum CEA, SUVmax and CT features, for predicting lymph node metastasis, the most important and significantly independent risk factors identified were SUVmax, CEA level, mediastinal or hilar lymphadenopathy, cavitation/bubble-likelucency and pleural retraction (P<0.05).Conclusion The lymph node metastasis is associated with SUVmax of primary tumor, serum CEA level, mediastinal or hilar lymphadenopathy, cavitation/bubble-likelucency and pleural retraction.The combination of radiographic features and serum CEA can help to predict more accurately the risk of lymph node metastasis in patients with peripheral lung adenocarcinoma.