OBJECTIVES: Weekly gemcitabin and erlotinib are both active in elderly patients treated for NSCLC. The aim of the GFPC0505 randomized phase II trial was to compare the efficacy and the cost of weekly gemcitabin (G) followed by erlotinib after progression (arm A) versus erlotinib followed by G after progression (arm B) in frail elderly patients with advanced non small-cell lung cancer (NSCLC), selected on the basis of a comprehensive geriatric assessment (CGA). METHODS: Frail elderly chemotherapy-naive patients with stage IIIB/IV NSCLC were selected after a CGA. Main clinical outcome was time to second progression (TTP2). Costs were limited to direct medical costs and were prospectively collected until progression, from the third party payer perspective. Health utilities (based on disease states and grade 3-4 toxicities) and costs after progression were derived from the literature. Sensitivity analyses were performed.  RESULTS: Median age of the 94 enrolled patients was 78.2 years, and 76 (80%) were male. There is no significantly difference between the 44 and 50 patients respectively randomized in arm A and B, in terms of efficacy (TTP2: 4.3 and 3.5 months: overall survival: 4.4 and 3.9 months, mean QALY:0.347 and 0.325) and in terms of mean direct costs (15,363 and 15,233€).  CONCLUSIONS: In this population, the 2 strategies appeared equivalent in terms of efficacy and costs. Supported by an unrestricted educational grant from Roche