This study investigated, for the D2 dopamine receptor, the relation between the ability of agonists and inverse agonists to stabilise different states of the receptor and their relative efficacies. Ki values for agonists were determined in competition versus the binding of the antagonist [3H]spiperone. Competition data were fitted best by a two-binding site model (with the exception of bromocriptine, for which a one-binding site model provided the best fit) and agonist affinities for the higher (Kh) (G protein-coupled) and lower affinity (Kl) (G protein-uncoupled) sites determined. Ki values for agonists were also determined in competition versus the binding of the agonist [3H]N-propylnorapomorphine (NPA) to provide a second estimate of Kh. Maximal agonist effects (Emax) and their potencies (EC50) were determined from concentration-response curves for agonist stimulation of guanosine-5'-O-(3-[32S]thiotriphosphate) ([35S]GTPgammaS) binding. The ability of agonists to stabilise the G protein-coupled state of the receptor (Kl/Kh determined from ligand-binding assays) did not correlate with either of two measures of relative efficacy (relative Emax, Kl/EC50) of agonists determined in [35S]GTPgammaS-binding assays, when the data for all of the compounds tested were analysed. For a subset of compounds, however, there was a relation between Kl/Kh and Emax. Competition-binding data versus [3H]spiperone and [3H]NPA for a range of inverse agonists were fitted best by a one-binding site model. Ki values for the inverse agonists tested were slightly lower in competition versus [3H]NPA compared to [3H]spiperone. These data do not provide support for the idea that inverse agonists act by binding preferentially to the ground state of the receptor.