Purpose: To devise a prognostic model based on clinical parameters for non-small cell lung cancer (NSCLC) patients treated with erlotinib as a salvage therapy. Patients and Methods: Between July 2006 and September 2008, two hundred fifty-seven metastatic/relapsed NSCLC patients who had been treated with erlotinib as a salvage therapy were analyzed retrospectively. Results: For the 257 patients, the median overall survival (OS) and progression-free survival (PFS) with erlotinib treatment were 12.4 and 2.8 months. Multivariate analysis showed that an ECOG performance status of 2 or more, an elevated serum LDH level, and the absence of skin rash were independent adverse prognostic factors for OS and that the presence of intra-abdominal metastasis, 2 or more prior chemotherapy regimens, and the absence of skin rash were prognostic factors for PFS. Patients were categorized into the following 4 prognosis groups on the basis of each adverse prognostic factor: good, intermediate, poor, and very poor prognosis. The median OS times for the good, intermediate, poor, and very poor prognosis groups were 22.0, 9.3, 5.4, and 2.7 months (p < 0.001) and the median PFS times were 6.5, 3.0, 1.2, and 0.9 months (p < 0.001). Conclusion: This prognostic model based on clinical parameters would be useful to identify patients who might be most likely to benefit from erlotinib therapy in clinical practice.