Clinical and experimental studies pertinent for demonstrating the antipsychotic potential of benzodiazepines (BDZ) and the involvement of γ-aminobutyric acid (GABA) in the origin of schizophrenia are reviewed. It is shown that, due to severe methodological problems and pitfalls, placebo-controlled, double-blind studies do not permit unequivocal conclusions on the efficacy of BDZs, but neither do they completely disprove it. Furthermore, at first glance, confusing and controversial findings in animal models indicate a bi-directionality of effects of full BDZ agonists on dopamine-mediated functions, which may perhaps be explained by (i) anatomical and functional organization of the GABA-dopamine system in the nigro-striatal and ventro tegmental area, and (ii) the regional non-selectivity of action of these drugs. The recent demonstration of structural polymorphism of the GABA(A)-BDZ receptor complex and heterogeneous distribution of sets of subunits of the GABA( A)-BDZ receptor in the brain, suggests possibilities for development of partial BDZ agonists showing greater regional selectivity of action and thus potentially more specific antipsychotic action. Initial clinical results with bretazenil (Ro 16-6028), a partial BDZ agonist, in acute schizophrenia are, in this respect, an encouraging lead to be followed further.