The concept that receptors can exist in multiple conformational states is becoming a physical reality. A fundamental question is how many active states need to be proposed in order to account for pharmacological observations, in particular, the finding that the same receptor type can exhibit a different agonist pharmacology when coupled to different effector pathways. In this article, Paul Leff, Clare Scaramellini, Clare Law and Ken McKechnie propose and develop a three-state receptor model in which two active conformations are assumed to exist. They show that this relatively simple theoretical model provides a basis for predicting variable agonist and inverse agonist behaviour in different systems containing the same receptor, and that it is able to account for emerging data obtained on promiscuously coupled receptors. It is argued that, while these new theoretical considerations challenge the fundamental assumptions and concepts of traditional receptor theory, the general principles of pharmacological receptor classification are largely preserved.