INTRODUCTION: Programmed cell death-1 (PD-1) inhibitors effectively treat NSCLC and prolong survival. Robust biomarkers for predicting clinical benefits of good response and long survival with anti-PD-1 therapy have yet to be identified; therefore, predictive biomarkers are needed to select patients with benefits. METHODS: We conducted a prospective study to explore whether serum antibody against NY-ESO-1 and/or XAGE1 cancer-testis antigens predicted primarily good clinical response and secondarily long survival with anti-PD-1 therapy for NSCLC. The serum antibody was detected by enzyme-linked immunosorbent assay, and tumor immune microenvironment and mutation burden were analyzed by immunohistochemistry and next-generation sequencing. RESULTS: In the discovery cohort (n = 13), six antibody-positive NSCLC cases responded to anti-PD-1 therapy (two complete and four partial responses), whereas seven antibody-negative NSCLC cases did not. Antibody positivity was associated with good response and survival, regardless of tumor programmed death ligand 1 (PD-L1) expression, mutation burden, and CD8+ T-cell infiltration. In the validation cohort (n = 75), 17 antibody-positive NSCLC cases responded well to anti-PD-1 therapy as compared with 58 negative NSCLC cases (objective response rate 65% versus 19%, p = 0.0006) and showed significantly prolonged progression-free survival and overall survival. Antibody titers highly correlated with tumor reduction rates. In the multivariate analysis, response biomarkers were tumor programmed death ligand 1 expression and antibody positivity, and only antibody positivity was a significantly better predictive biomarker of progression-free survival (hazard ratio = 0.4, p = 0.01) and overall survival (hazard ratio = 0.2, p = 0.004). CONCLUSIONS: Our results suggest that NY-ESO-1 and/or XAGE1 serum antibodies are useful biomarkers for predicting clinical benefits in anti-PD-1 therapy for NSCLC and probably for other cancers.