This meeting has well illustrated the state-of-the-art phenomenon of receptor-receptor interactions between heptaspanning membrane receptors (also named G protein-coupled receptors [GPCRs]), as both theoretical aspects and new experimental evidence have been discussed extensively.From a general standpoint the theoretical discussion has enlightened us to the phenomenon of receptor-receptor interactions as a "privileged window," which allows the exploration of some important structural and functional aspects of molecular networks (Agnati et al., this issue). Also, unexpected functions of well-known proteins can be studied, such as beta-arrestins, which are at the interface between horizontal and vertical molecular networks and have a signaling function of their own besides modulating receptor transduction and trafficking (Lefkowitz, this issue). The experimental data have shown that many in vitro approaches are now available to study the structure and function of homo- and hetero-oligomers in cell lines and primary cultures, making possible a major expansion of the field. However, the in vivo approaches to study the role of heteromeric receptor complexes in cell function have been less clear cut. It was discussed, especially by George, that one problem with receptor knockout mice is inter alia that the removal of one receptor will alter the balance of various homomeric and heteromeric receptor complexes in which the deleted receptor participated. In fact, each receptor might participate in a number of heteromeric receptor complexes. An attractive in vivo approach in the receptor-receptor interaction study will be to use knockin mice in which one receptor A (Ra) has been removed and replaced by a mutated A receptor (Ra'). Several different strategies can be considered, and here are some of them. 1. Ra' can still recognize its ligand and signal but does not interact with its usual heteromeric receptor partner B (Rb). Thus, Ra' operates as a monomer and/or homomer. 2. Ra' can still recognize its ligand and interact with Rb, but it cannot signal (e.g., it does not couple to the G protein). Thus, Ra' operates as a modulator protein of Rb. 3. Ra' can still recognize its ligand and signal, but it allosterically affects its heteromeric partner Rb in a different way. Thus, Ra' operates as an unusual modulator of Rb.