OBJECTIVES: Although several clinical and biological parameters are prognostic factors of NSCLC patients outcome, their medico-economic impact in the prescription of erlotinib has never been evaluated. A French NCI prospective study aimed to compare cost and effectiveness of three strategies of erlotinib initiation in second line or more treatment of advanced NSCLC patients: initiation in all patients, patients selected on clinical-guided strategy and patients selected on biological-guided strategy. METHODS: A Markov model compared the outcomes and costs (limited to direct medical costs from the third party payer perspective) of a prospective multicentric cohort of consecutive advanced NSCLC patients newly treated by erlotinib, to a cohort of clinical-selected patients (non/ex-smoking women with adenocarcinoma histology) and a cohort of biomarker-selected patients (EGFR mutation). Utility data were extracted from literature. Sensitivity analyses were performed. RESULTS: A total of 522 patients were enrolled between March 2007 and March 2008. Median age was 63 years; 32% were females; 65% had adenocarcinoma and 8% had EGFR mutation. The strategy which consists to treat all patients was dominated, as it was both the less effective and the most expensive strategy (0.495 QALY/€22,396). The clinical-guided strategy was slightly more effective than the biological-guided strategy (respectively 0.568 and 0.563 QALY), but it was also more expensive (respectively €16,299 and €15,187). The dominant strategy was then the biological-guided strategy (€26,975/QALY). The model was robust to variations of biological exam costs, palliative costs and utility data. Biological-guided strategy appears the most effective and the less expensive strategy when the prevalence of EGFR mutation exceeds 10%. CONCLUSIONS: Biological-guided strategy appears the dominant strategy if the prevalence of EGFR mutation was > 10%. This suggests determining EGFR mutation status in priority to non/former smokers, females with adenocarcinoma.