For two decades it has been impossible to develop drugs with novel mechanisms of action against herpesviruses, and treatment has been confined largely to the use of inhibitors of viral DNA polymerase. As a representative of a novel inhibitory approach, the non-nucleosidic BAY 38-4766 was identified as a highly selective inhibitor of human cytomegalovirus (HCMV). The compound selectively inhibits not only HCMV strains, including ganciclovir-resistant, ganciclovir/foscarnet and ganciclovir/cidofovir double-resistant clinical isolates, but also a number of monkey and rodent cytomegaloviruses. In a murine cytomegalovirus (MCMV) pathogenicity model in mice, antiviral efficacy and excellent tolerability were demonstrated. BAY 38-4766-resistant HCMV and MCMV strains are not cross-resistant to the nucleoside analogues ganciclovir and cidofovir or the pyrophosphate analogue foscarnet, indicating a different mode of action. Mechanistic studies demonstrated that the high selectivity of this drug class is most likely due to the inhibition of a late stage of the viral replication cycle. Sequence analyses of resistant HCMV and MCMV strains revealed mutations in UL89 and UL104, proteins known to be involved in viral DNA cleavage and packaging. Consequently, the drug is highly specific for the viral as opposed to cellular functions, since UL89 is related to a bacteriophage terminase and no human equivalent exists. In addition, because some of the genes of the viral DNA cleavage and packaging complex are highly conserved among herpesviruses, development of broad-spectrum agents covering additional human herpesviruses might be possible using this approach.