Epidermal growth factor receptor tyrosine kinase inhibitors (EGFR-TKIs) demonstrated to significantly improve survival of non-small cell lung cancer patients (NSCLC) previously exposed to chemotherapy. Although clinical features, particularly smoking history, help physicians for identifying the sensitive population, a proper patient selection should not preclude to drug target assessment. EGFR mutations or increased EGFR gene copy number assessed by fluorescence in situ hybridization (FISH) identify NSCLC with the highest chance to respond to the therapy. Although indirect comparisons suggest that mutation analysis is the best available technique for identification of responders, survival improvement is not confined to individuals with tumor shrinkage. For patients with metastatic NSCLC, where definitive cure in not achievable, response is probably not the best end-point, since survival improvement observed with TKI included also patients with stable or progressive disease. Data from large randomized studies indicated that FISH technology is probably the best method for patient selection when the main end-point is survival. FISH was the only EGFR test significantly associated with prolonged survival in large randomized trials with a control arm of placebo, the only studies able to discriminate between predictive and prognostic value of such biomarkers. Moreover, in absence of any convincing data on the prognostic role of EGFR FISH or EGFR mutations, results from large phase III trials suggest that patients with clinical or biological predictors for TKI sensitivity survive longer when exposed to standard chemotherapy, a relevant aspect that should be considered in designing clinical trials.