G-protein-coupled receptors (GPCRs) are, and will probably remain, the most tractable class of targets for the development of small-molecule therapeutic medicines. Currently, all approved GPCR-directed medicines are agonists or antagonists at orthosteric binding sites - except for the calcimimetic cinacalcet, which is a positive allosteric modulator of Ca(2+)-sensing receptors, and maraviroc, an allosteric inhibitor of CC-chemokine receptor (CCR) 5. It is now widely accepted that GPCRs exist and might function as dimers, and there is growing evidence for the physiological presence and relevance of GPCR heterodimers. Molecules that can regulate a GPCR within a heterodimer, through allosteric effects between the two protomers of the dimer or between a protomer or protomers and the associated G protein, offer the potential to function in a highly selective and tissue-specific way. Despite the conceptual attraction of such allosteric regulators of GPCR heterodimers as drugs, they cannot be identified by screening approaches that routinely use a 'one GPCR target at a time' strategy. In our opinion, this will require the development of new approaches for screening and a return to the use of physiologically relevant cell systems at an early stage in compound identification.