The hypothesis that the cognitive decline in senile dementia is related to the loss of cortical cholinergic afferent projections predicts that pharmacological manipulations of the remaining cholinergic neurons will have therapeutic effects. However, treatment with cholinesterase inhibitors or muscarinic agonists has been, for the most part, largely unproductive. These drugs seem to disrupt the normal patterning of cholinergic transmission and thus may block proper signal processing. An alternative pharmacological strategy which focuses on the amplification of presynaptic activity without disrupting the normal patterning of cholinergic transmission appears to be more promising. Such a strategy may make use of the normal GABAergic innervation of basal forebrain cholinergic neurons in general, and in particular of the inhibitory hyperinnervation of remaining cholinergic neurons which may develop under pathological conditions. Disinhibition of the GABAergic control of cholinergic activity is assumed to intensify presynaptic cortical cholinergic activity and to enhance cognitive processing. Although the extent to which compounds such as the benzodiazepine receptor antagonist beta-carboline ZK 93,426 act via the basal forebrain GABA-cholinergic link is not yet clear, the available data suggest that the beneficial behavioral effects of this compound established in animals and humans are based on indirect cholinomimetic mechanisms. It is proposed that an activation of residual basal forebrain cholinergic neurons can be achieved most physiologically via inhibitory modulation of afferent GABAergic transmission. This modulation may have a therapeutic value in treating behavioral syndromes associated with cortical cholinergic denervation.