Most drugs acting on G-protein-coupled receptors (GPCRs) are classically defined as agonists, partial agonists or antagonists. This simplified classification seems sufficient to explain most of their therapeutic properties. The more recent description of inverse agonism has helped to revise theoretical models of GPCR function, but the therapeutic implications of the new concepts remain clearly restricted. Further complexity has arisen with demonstrations that a given receptor can adopt various conformations that support coupling with distinct G proteins. Because the related signaling pathways seem to be differentially affected by some ligands, the concept of 'functional selectivity' has been proposed, calling for a revision of the definitions of agonism and intrinsic efficacy. Evidence of complexity in G-protein coupling and examples of functional selectivity are accumulating, opening perspectives for drug development. Although such complexity should be regarded as an opportunity to gain pharmacological specificity, unraveling the physiological implications of these concepts is essential before their therapeutic relevance can be defined.