Evidence is presented which indicates that the central actions of the benzodiazepines cannot be fully accounted for by assuming an action only at the GABAA-Cl- channel supramolecular complex. 2. The hypothesis is presented, together with supporting evidence, that inhibition of adenosine uptake can account for many of the actions of the benzodiazepines. 3. New findings showing that Ro 15-1788 and Ro 5-4864 have both potentiative and antagonistic interactions with adenosine are discussed. 4. The proconvulsant beta-carbolines are shown to be adenosine antagonists. 5. The concept that benzodiazepine action may involve several mechanisms is presented.