The purpose of these studies were threefold. Firstly, to further characterize the effect of perforant path transection on a test of short-term memory: delayed matching (or nonmatching)-to-position [D(N)MTP]. Secondly, to evaluate the effect of a transient cerebral ischaemia in the same task. Both surgical procedures were chosen as they produce a CNS lesion similar to that described in Alzheimer's Disease (AD). Thirdly, the effect of the acetylcholinesterase inhibitor, donepezil (Aricept(R), E2020), on the resulting cognitive impairment was studied. Perforant path transection produced a robust, delay-dependent impairment of choice accuracy in rats performing either a delayed matching- or nonmatching-to-position task. Sample latency was also reduced following lesion, yet the lesion-induced impairment was not affected by increasing the response requirement at the sample stage. An 11-min period of transient ischaemia (two-vessel occlusion model) resulted in almost complete loss of hippocampal CA1 pyramidal cells and a delay-dependent impairment in DMTP performance. However, unlike perforant path lesions, this deficit was unstable and declined in magnitude over the experimental period. Increasing the delay interval restored this deficit. Donepezil, at doses that robustly attenuated a scopolamine (0.06 mg/kg s.c.)-induced DMTP accuracy impairment in naïve, unoperated rats, had no effect against either lesion-induced impairment. The results are considered in terms of the effectiveness of acetylcholinesterase inhibitors in noncholinergic-based preclinical cognitive models.