The full-length genome of human cytomegalovirus strain AD169 was cloned as an infectious bacterial artificial chromosome (BAC) plasmid, pAD/Cre. The BAC vector, flanked by LoxP sites, was inserted immediately after the Us28 open reading frame without deletion of any viral sequences. The BAC vector contained the Cre recombinase-encoding gene disrupted by an intron under control of the simian virus 40 early promoter. When pAD/Cre was transfected into primary human foreskin fibroblast cells, Cre was expressed and mediated site-specific recombination between the two LoxP sites, excising the BAC DNA backbone. This gave rise to progeny virus that was wild type with the exception of an inserted 34-bp LoxP site. We performed site-directed mutagenesis on pAD/Cre to generate a series of viruses in which the TRL/IRL13 diploid genes were disrupted and subsequently repaired. The mutants reach the same titer as the wild-type virus, indicating that the TRL/IRL13 open reading frames are not required for virus growth in cell culture. The sequence of the TRL13 open reading frame in the low-passage Toledo strain of human cytomegalovirus is quite different from the corresponding region in the AD169 strain. One of multiple changes is a frameshift mutation. As a consequence, strain Toledo encodes a putative TRL13 protein whose C-terminal domain is larger (extending through the TRL14 coding region) and encodes in a reading frame different from that of strain AD169. We speculate that the strain AD169 coding region has drifted during passage in the laboratory. We propose that TRL13 has been truncated in strain AD169 and that the partially overlapping TRL14 open reading frame is not functional. This view is consistent with the presence of both TRL13 and -14 on all mRNAs that we have mapped from this region, an organization that would include the much longer strain Toledo TRL13 open reading frame on the mRNAs.