The current status of the cholinergic hypothesis of cognitive dysfunction in Alzheimer's disease is reviewed in the context of recent attempts to alleviate specific cognitive impairments produced in rats by excitotoxic lesions of basal forebrain neurons by treatment with cholinergic agents. AMPA-induced lesions of the nucleus basalis region in rats produce profound and relatively specific reductions in neocortical markers of cholinergic function but fail to affect performance in many tests of memory and learning in rats. However, such lesions produce specific deficits in responding accurately in a test of visual attentional performance, which are reversed dose-dependently by treatment with systemic physostigmine or nicotine. Analogous improvements have been reported in a clinical trial of the anticholinesterase tacrine in patients with Alzheimer's disease. By contrast, AMPA-induced lesions of the medial septum produce profound reductions in hippocampal acetylcholine and accompanying delay-dependent deficits in a delayed non-matching-to-position procedure which measures spatial working memory in rats. This impairment is shown to be reversed to some extent by treatment with low doses of physostigmine. The results are discussed in terms of the multivariate nature of the neurochemical pathology of Alzheimer's disease and attendant limitations in the use of the cholinergic strategy. The cognitive costs, as well as benefits, of cognitive enhancers are discussed, as well as the need to broaden our therapeutic approach to other neurotransmitter systems and other neurodegenerative disorders.