8061 Background: Pem has been used for pts with previously treated advanced NSCLC. We tried to examine whether the response to prior systemic therapy regimen could predict the efficacy of subsequent Pem therapy. Methods: The medical record of clinical stage IIIB or IV NSCLC pts who received Pem as the second-line or further-line treatment were analyzed retrospectively. Prior systemic therapies were divided into 4 types [gemcitabine based (G), paclitaxel based (P), docetaxel based (D), and EGFR tyrosine kinase inhibitors (I)]. Along with that, patients were classified into two response groups, either responders (partial responses or stable disease for 4 months or more) or non-responders to each type of the prior therapy. Response rate (RR) and progression free survival (PFS) for Pem therapy were analyzed according to the response groups for each type of prior therapies. Results: A total of 247 pts received Pem therapy, and their median PFS was 2.3 months. The number of pts who previously received G, P, D and I, was 159, 120, 110, and 139, respectively. The RR for Pem was higher in responders to G therapy than in non-responders to G (15.0% vs 4.3%, p = 0.02). In addition, median PFS after Pem therapy was longer in responders to G therapy than in non-responders (3.0 vs 1.7 months, p = 0.007). However, the responses to prior P, D, I therapies had no impact on the efficacy of subsequent Pem therapy. By univariate analyses, the variables of the responders to G therapy, female, never-smoker, ECOG performance status 0–1 were good predictive factors for Pem therapy in terms of PFS. By multivariate analysis, only the responders to G therapy had a statistical significance (Hazard ratio = 0.55; 95% CI, 0.37–0.82). Conclusions: The response to the prior gemcitabine based therapy was a predictive factor for subsequent pemetrexed therapy for advanced NSCLC. No significant financial relationships to disclose.