Formation of the glycoprotein gH/gL heterooligomer has important implications for understanding the pathology of human herpesvirus-6(HHV-6)-associated disease because this complex is essential for infectivity and fusogenic cell-to-cell spread. Definition of the HHV-6 gH domain involved in protein-protein interactions was addressed by targeting regions defined by conserved cysteines identified by alignment of gH amino acid sequences representative of all herpesvirus subfamilies. Studies using site-directed mutagenesis and transient cellular expression showed that the N-terminus of HHV-6 gH includes a 230-amino-acid domain required for interaction with HHV-6 gL encompassing residues conserved specifically amongst betaherpesviruses. Interestingly, the human cytomegalovirus (HCMV) homologues, UL75 (gH) or UL115 (gL), can substitute for HHV-6 glycoproteins and participate in heterologous complex formation. Furthermore, the region which governs this heterologous gL binding also maps to the N-terminal portion of HHV-6 gH. Although both proteins can functionally substitute for complex formation there are also specific differences. Surprisingly, further deletion of HHV-6 gH to 145-amino-acid-domain residues abolishes complex formation with HHV-6 gL but allows interaction with HCMV gL. This may be related to requirements in HHV-6 for homodimer formation before complex formation between gH and gL. Under nonreducing conditions HHV-6 gH and gL form multimeric complexes consistent with intra- and intermolecular dimer formation stabilised by disulphide bonds whereas for HCMV there is no evidence for dimer formation for gH and multimeric complexes have only been observed between gH and gL. In summary, both HHV-6 and HCMV glycoproteins can interact and the heterologous complex between HHV-6 gH and HCMV gL is possibly more stable. This may result in important biological consequences in vivo during cellular coinfections by facilitating spread of the viruses, with applications to altered cellular tropisms and effects on reactivation from the latently infected cell.