Purpose Persistently elevated post-treatment plasma EBV DNA is a robust predictor of relapse in NPC. However, assay standardization is necessary for use in biomarker-driven trials. We conducted a study to harmonize the method between four centers with expertise in EBV DNA quantitation. Experimental Design Plasma of 40 NPC patients were distributed to four centers. DNA was extracted and EBV DNA copy number was determined by real-time quantitative PCR (BamHI-W primer/probe). Centers used the same protocol but generated their own calibrators. A harmonization study was then conducted using the same calibrators and PCR master mix and validated with ten pooled samples. Results The initial intraclass correlations (ICC) for the first 40 samples between each center and the index center were 0.62 (95%CI: 0.39–0.78), 0.70 (0.50–0.83) and 0.59 (0.35–0.76). The largest variability was the use of different PCR master mixes and calibrators. Standardization improved ICC to 0.83 (0.5–0.95), 0.95 (0.83–0.99) and 0.96 (0.86–0.99), respectively, for ten archival frozen samples. For fresh plasma with spiked-in EBV DNA, correlations were >0.99 between the centers. At five EBV DNA copies/reaction or above, the coefficient of variance (CV) was <10% for the cycle threshold (Ct) among all centers, suggesting this concentration can be reliably used as a cutoff for defining the presence of detectable EBV DNA. Conclusions Quantitative PCR assays, even when performed in experienced clinical labs, can yield large variability in plasma EBV DNA copy numbers without harmonization. The use of common calibrators and PCR master mix can help to reduce variability.