The cloned human beta-3 adrenoceptor was expressed in Chinese hamster ovary cells at three different levels (130, 400 and 3000 fmol/mg). The potency and intrinsic activity of a range of agonists in functional assays with these cell lines rose as a function of increasing receptor density. Operational analysis of concentration-response data allowed calculation of functional affinity and efficacy of agonists at the human beta-3 adrenoceptor. The data highlighted the low efficacy of BRL 37344 ¿(RR,SS)-(+/-)-4-[(2-(2-(3-chlorophenyl)-2-hydroxyethyl)amino)-propyl] phenoxyacetate¿ for the human beta-3 adrenoceptor, which may explain its lower potency at the human receptor despite its higher affinity relative to isoprenaline. The potency of catecholamines at the human beta-3 adrenoceptor was found to be 1 to 2 orders of magnitude higher when determined in an intact cell cAMP accumulation assay compared with a membrane-based adenylyl cyclase activation assay. The reason for this enhanced sensitivity is not clear, but the result is that the potency of the natural agonist noradrenaline in the intact cell is considerably higher than predicted either from its ligand binding affinity, or from its potency in membrane-based assays. Much smaller enhancements in sensitivity were observed for compounds of the aryloxypropanolamine class such as CGP 12177 [(+/-)-4-(3-t-butylamino-2-hydroxypropoxy)benzimidazol-2-one], with the result that the rank order of potency of such agonists at the beta-3 adrenoceptor was altered. In particular, CGP 12177 exhibited high relative potency in the cyclase assay, but low relative potency in intact cell assays. These findings highlight the importance of selecting appropriate expression levels and appropriate assay methodology when cloned receptors are used to characterize agonists.