Background: Several studies have investigated the feasibility of detecting EGFR mutations in liquid biopsies of Non-Small Cell Lung Cancer (NSCLC) patients. However, the potential to accurately quantify EGFR mutations in plasma for clinical purposes is largely unexplored. Methods: Plasma samples were obtained from 79 subjects: (a) 42 NSCLC patients with EGFR mutations in primary tumors tissue, from a prospective trial, where blood samples were collected at baseline prior to first-line erlotinib therapy and immediately after progression; (b) 15 previously untreated stage IIIB-IV NSCLC patients, with EGFR mutation positive tumor specimens, where plasma samples were collected at baseline and serially at 4-60 days during TKI therapy; (c) 22 negative control cases. EGFR mutation analysis in plasma was conducted by the cobas EGFR Mutation Test (EGFR test, under development, RMS, Pleasanton, CA) and ultra-deep next generation sequencing (UDNGS) by Roche 454-GS Junior and Illumina MiSeq. A semi-quantitative index (SQI) was derived from a dilution series of known mutation copy numbers. Clinical response, expressed as percent tumor shrinkage (PTS), was evaluated according to RECIST criteria. Results: The sensitivity and specificity of the EGFR test and UDNGS assay in plasma versus tissue was 72% and 100%, and 74% and 100%, respectively. Quantitative indices by the EGFR test and UDNGS showed a significant positive correlation (p < 0.0001). Serial testing of EGFR mutations revealed a progressive decrease in the EGFR semi-quantitative index (SQI) during therapy in all of the patients, starting from the 4^thday of treatment in 90% of cases. The rate of SQI decrease was more than 50% at 15 days in 73% of patients (rapid responders) and less than 50% at 15 days in 27% of patients (slow responders) and was correlated with PTS at 2 months. Conclusions: Quantification of EGFR mutations with the cobas test in plasma is feasible. The variation of EGFR SQI during therapy could be useful for early prediction of response and diagnosis of relapse, with further implications for patient management. A prospective trial has been planned to confirm these results. Clinical trial information: EudraCT Number: 2010-023892-24.