Background: In patients with activating EGFR mutations and ALK fusion, target specific tyrosine kinase inhibitor (TKI) showed significant survival improvement compared to the cytotoxic chemotherapy. However, the questions remain which combination strategy will be the best option for the patients who have failed from TKI. Especially, the role of an immune checkpoint inhibitor (ICI) in this population is still unclear. This study is designed and conducted based on the recent subgroup analyses from the IMpower 150 study which showed the positive clinical outcomes of atezolizumab combined with VEGF inhibitor and conventional cytotoxic chemotherapy in EGFR mutation and ALK translocation. Methods: This study is the phase III, open-label, multicenter study of atezolizumab in combination with bevacizumab + carboplatin + paclitaxel (ABCP, Arm A) compared with pemetrexed + cisplatin or carboplatin (Arm B). The study population will be randomized to either Arm A (n = 152) or Arm B (n = 72) based on two stratification factors, EGFR vs. ALK and presence of brain metastases. In Arm A, patients will be treated with 4 or 6 cycles of ABCP followed by maintenance atezolizumab and bevacizumab every three weeks. In Arm B, pemetrexed maintenance therapy will be applied every three weeks after 4 or 6 cycles of pemetrexed + cisplatin or carboplatin. As key inclusion criteria, the patients must be diagnosed with stage IV non-squamous non-small cell lung cancer with either activating EGFR mutation or ALK translocation. All the patients need to be cytotoxic chemotherapy naïve and must have experienced disease progression to treatment with at least one EGFR or ALK TKI. If the patients have T790M mutation after 1^st or 2^nd generation EGFR TKI, second line 3^rd generation EGFR TKI treatment is mandatory. The number of T790M positive patients is restricted to under 30% of the entire study population. The primary endpoint is progression-free survival and the major secondary endpoints are overall survival, objective response rate and duration of response. A total of 228 subjects will be enrolled to detect a hazard ratio of 0.67. The first subject received treatment in Aug. 2019 and 19 patients receive the treatment. This study is opened in 3 sites and expected to be opened at 18 sites in South Korea. The time point for the primary analyses is Q3. 2022. Clinical trial information: NCT03991403.