Epstein-Barr virus (EBV) replicates its genome as a licensed plasmid in latently infected cells. Although replication of this plasmid is essential for EBV latent infection, its synthesis still fails for 16% of the templates in S phase. In order to understand these failures, we sought to determine whether the affinity of the initiator protein (EBNA1) for its binding sites in the origin affects the efficiency of plasmid replication. We have answered this question by using several engineered origins modeled upon the arrangement of EBNA1-binding sites found in DS, the major plasmid origin of EBV. The human TRF2 protein also binds to half-sites in DS and increases EBNA1's affinity for its own sites; we therefore also tested origin efficiency in the presence or absence of these sites. We have found that if TRF2-half-binding sites are present, the efficiency of supporting the initiation of DNA synthesis and of establishing a plasmid bearing that origin directly correlates with the affinity of EBNA1 for that origin. Moreover, the presence of TRF2-half-binding sites also increases the average level of EBNA1 and ORC2 bound to those origins in vivo, as measured by chromatin immunoprecipitation. Lastly, we have created an origin of DNA synthesis from high-affinity EBNA1-binding sites and TRF2-half-binding sites that functions severalfold more efficiently than does DS. This finding indicates that EBV has selected a submaximally efficient origin of DNA synthesis for the latent phase of its life cycle. This enhanced origin could be used practically in human gene vectors to improve their efficiency in therapy and basic research.