Membrane chloride (Cl-) permeability was studied in a novel subcellular brain preparation, the synaptoneurosome. Using a radioactive tracer exchange technique, Cl- transport was determined by measuring 36Cl- efflux from rat cerebral cortical synaptoneurosomes. Barbiturates increased 36Cl- efflux in a dose-dependent manner with the following relative order of potency: 5-(1,3-dimethylbutyl)-5-ethyl barbituric acid ((-)-DMBB) greater than pentobarbital greater than secobarbital greater than (+)-DMBB greater than hexobarbital greater than amobarbital greater than mephobarbital. Phenobarbital and barbital were virtually inactive. A good correlation was observed between the potencies of these barbiturates in stimulating 36Cl- efflux and their anesthetic potencies in mice (r = 0.90, p less than 0.01) and their abilities to enhance [3H] diazepam binding to brain membranes (r = 0.77, p less than 0.05). The effect of pentobarbital in enhancing 36Cl- efflux was reversed by the gamma-aminobutyric acid (GABA) antagonists picrotoxin and bicuculline. Picrotoxin and bicuculline both decreased 36Cl- efflux in the absence of pentobarbital, suggesting the presence of endogenous GABA. Incubation of synaptoneurosomes with 4,4'-di-isothiocyano- or dinitro-2,2'-disulfonic acid stilbene, inhibitors of anion transport, also decreased both basal and pentobarbital-induced 36Cl- efflux. Pentobarbital (500 microM) was most effective in inducing 36Cl- efflux in the cerebellum, hippocampus, and cortex (23.7, 23.6, and 22.5%, respectively), and was less effective in stimulating 36Cl- efflux in the striatum (15.1%) and pons-medulla (6.2%). The relative efficacy of pentobarbital in enhancing 36Cl- efflux among these various brain regions was highly correlated (r = 0.96, p 0.01) with the relative densities of [35S]-t-butylbicyclophosphorothionate-binding sites, a measure of GABA-gated Cl- channel density.(ABSTRACT TRUNCATED AT 250 WORDS)