There has recently been a large increase in the number of new benzodiazepine receptor ligands, some with benzodiazepine structures, but many with other chemical structures. The pharmacological activities of these ligands have been interpreted on the basis of a continuum of efficacy ranging from full agonist through different degrees of partial agonism to antagonist and through partial to full inverse agonists. Inconsistencies with this hypothesis are considered in terms of alternative hypotheses, particularly the existence of functionally separable receptor subtypes. The potential of partial agonists as non-sedative anxiolytic agents with reduced potential of dependence and of weak partial inverse agonists as pro-cognitive agents is discussed. A pharmacophore for benzodiazepine receptors is proposed and supporting evidence presented.