The binding of [3H]GABA to membranes prepared from human brains obtained post morten was examined. This binding was independent of patient sex, age (16--80 years), postmortem interval (4--33 h) or storage time when frozen (0-64 months). In preparations from cerebellar cortex various compounds displaced [3H]GABA binding with the following order of potency: muscimol greater than 3-aminopropanesulfonic acid greater than GABA greater than imidazoleacet acid greater than delta-amino-n-valeric acid greater than 3-hydroxyGABA greater than bicuculline. Other compounds active 'in vitro' included strychnine, homocarnosine and some (e.g. clozapine, thioridazine, pimozide) but not all (chlorpromazine, haloperiodol) neuroleptics. Compounds inactive 'in vitro' included aminooxyacetic acid, baclofen, picrotoxin, anticholinergics, metrazole, anticonvulsants and naloxone. Triton X-100 augmented the [3H]GABA binding (25 nM) by about 10--20-fold in most brain regions. [3H]GABA binding (IC50) was altered in Huntington's chorea and Reye's syndrome, but not in schizophrenics (4-neuroleptic-treated patients) or sudden infant death syndrome. The data presented strongly support the proposal that the measurement of [3H]GABA binding in postmortem human brain offers a reflection of the state of the physiologically relevant GABA receptor.