Tumor-infiltrating immune cells are closely associated with clinical outcome. However, immunohistochemistry-based analysis of tumor infiltrates can be misleading as the representative marker of an immune subpopulation might be expressed in other cell types. In this study, based on a metagene approach (known as CIBERSORT) and an online databse, The Cancer Immunome Atlas (tcia.at/), we comprehensively analyzed the tumor-infiltrating immune cells present in lung adenocarcinoma (LUAD) and lung squamous cell carcinoma (LUSC). A total of 22 types of both adaptive and innate tumor-infiltrating immune cells were evaluated in LUAD (n=492) and LUSC (n=488). As a result, tumors lacking memory B cells or with increased number of M0 macrophages were associated with the poor prognosis in LUAD at early clinical stage. In LUSC, T follicular helper cells were associated with favorable outcome, while increased number of neutrophils predicted a poor outcome. Moreover, Kaplan-Meier analysis of the prognostic value of immune checkpoint molecules revealed that expression of ICOS was positively correlated the clinical outcome of patients with LUAD. Collectively, our data suggest that tumor-infiltrating immune cells in lung cancer are likely to be important determinants of both prognosis and response to immunotherapies.