Objective Digital PCR ( dPCR ) was established to detect plasma epidermal growth factor receptor (EGFR) T790M mutation of non-small cell lung cancer (NSCLC) patients and was evaluated in terms of analytical performance and clinical application significance.Methods The specific primers and probes for EGFR T790M mutation and wildtype were designed to establish dPCR platform.Limit of blank, sensitivity and linearity of dPCR were evaluated by the detection of plasmids with different concentrations to set up optimal reporting system and reanalyzing process.The mutation of EGFR T790M in plasma and tissue samples from 10 patients with advanced NSCLC resistant to EGFR-TKI therapy who were enrolled in Zhongshan Hospital Fudan University from January 2014 to October 2015 were analyzed by dPCR and amplification refractory ( ARMS) , respectively.The consistency was evaluated between dPCR and ARMS by Chi-square test.The correlation of T790M abundance detected by dPCR between plasma and tissue samples was also analyzed by Peasrson correlation analysis.Results Limit of blank and sensitivity of dPCR was 10 copies and 0.01%, respectively.dPCR was evaluated as linear in the range of 0.01%-100%( Y=1.226X-3.984,R2 =0.999 ).The consistency between dPCR and ARMS of tissue samples was good ( kappa=0.80), while the positive rates of plasma T790M detected by dPCR was significantly higher than ARMS (50%vs 20%,P<0.05).It was found that T790M abundance detected by dPCR was highly correlated between lung cancer tissue and plasma ( R =0.923, P <0.05 ) using Pearson correlation analysis. Conclusions A new method of dPCR with high sensitivity and absolute quantification is established for the detection of EGFR T790M mutation in plasma from advanced NSCLC patients, which brings tumor liquid biopsy into real.It has the ability to provide the most direct and valuable guidance for clinicians to make decision on EGFR tyrosine kinase inhibitors therapy in patients with advanced NSCLC resistant to EGFR-TKI.