MRK-409 binds to α1-, α2-, α3- and α5-containing human recombinant GABA(A) receptors with comparable high affinity (0.21-0.40 nM). However, MRK-409 has greater agonist efficacy at the α3 compared with α1 subtypes (respective efficacies relative to the full agonist chlordiazepoxide of 0.45 and 0.18). This compound readily penetrates the brain in rats and occupies the benzodiazepine site of GABA(A) receptors, measured using an in vivo [(3)H]flumazenil binding assay, with an Occ(50) of 2.2 mg/kg p.o. and a corresponding plasma EC(50) of 115 ng/mL. Behaviourally, the α3-preferring agonist efficacy profile of MRK-409 produced anxiolytic-like activity in rodent and primate unconditioned and conditioned models of anxiety with minimum effective doses corresponding to occupancies, depending on the particular model, ranging from ∼35% to 65% yet there were minimal overt signs of sedation at occupancies greater than 90%. In humans, however, safety and tolerability studies showed that there was pronounced sedation at a dose of 2 mg, resulting in a maximal tolerated dose of 1 mg. This 2 mg dose corresponded to a C(max) plasma concentration of 28 ng/mL, which, based on the rodent plasma EC(50) for occupancy of 115 ng/mL, suggested that sedation in humans occurs at low levels of occupancy. This was confirmed in human positron emission tomography studies, in which [(11)C]flumazenil uptake following a single dose of 1 mg MRK-409 was comparable to that of placebo, indicating that occupancy of GABA(A) receptor benzodiazepine binding sites by MRK-409 was below the limits of detection (i.e. <10%). Taken together, these data show that MRK-409 causes sedation in humans at a dose (2 mg) corresponding to levels of occupancy considerably less than those predicted from rodent models to be required for anxiolytic efficacy (∼35-65%). Thus, the preclinical non-sedating anxiolytic profile of MRK-409 did not translate into humans and further development of this compound was halted.