Categorizations of best response observed at week 8 (between week 3 and 14) of first-line treatment in two studies of bevacizumab plus chemotherapy in Western (878 patients) and Chinese (198 patients) patients with non-small cell lung cancer were assessed together with baseline prognostic factors in multivariate parametric models to predict overall survival (OS) and progression free survival (PFS). Predictive performances of the models were assessed by simulating multiple replicates of the studies. Disease control rate (DCR) was the best response categorization to predict OS and PFS. In the OS model, DCR fully captured bevacizumab effect. For PFS, DCR did not fully capture bevacizumab treatment effect. The models adequately predicted OS and PFS distributions in each arm as well as bevacizumab hazard ratio (HR) for OS and PFS, for example, in Western patients (model prediction [95% prediction interval]: 0.84 [0.71–0.98] vs. observed: 0.77 for OS and 0.59 [0.49–0.72] vs. observed: 0.58 for PFS). Covariates in the models captured endpoint differences seen in Chinese patients. There was no impact of Chinese ethnicity on the DCR relationship to OS or PFS. DCR predicted OS benefit with bevacizumab in first-line NSCLC patients. Western data can be used to inform design of studies in Chinese patients.