Two experiments examined the effects of excitotoxic lesions of the substantia innominata on cholinergic activity in the neocortex and on performance in a paradigm measuring selective attention in the rat. In Expt. 1, ibotenate-induced lesions produced approximately 30% reductions in cortical choline acetyltransferase (ChAT) activity, and damage to wide regions of the substantia innominata and ventral pallidum. The rats were impaired in their ability to localize brief visual targets in a serial reaction time task, as measured by reduced choice accuracy. This impairment was particularly evident at short stimulus durations, but the lesioned rats did not exhibit evidence of primary visual sensory dysfunction and exhibited only minor deficits when the stimuli were presented unpredictably. The deficit was exacerbated when distracting white noise was interpolated into the task. The rats with lesions were also slower to respond correctly, probably resulting partly from the adoption of a speed/error trade-off strategy, and were slower to collect earned food pellets, although they made no more errors of omission than controls. In Expt. 2, quisqualate-induced lesions produced fewer signs of non-specific damage and 50% reductions in cortical ChAT activity. This lesion produced generally qualitatively similar, but weaker effects to those of ibotenate-induced lesions. It was notable that many of the deficits following either ibotenate- or quisqualate-induced lesions lasted for several months after surgery. The results are discussed in terms of the cholinergic hypothesis of cognitive dysfunction. It is argued that lesions of the substantia innominata, including the magnocellular cholinergic neurons of the nucleus basalis of Meynert, produce deficits in attentional processing, which may not result from damage specifically to cholinergic cells. However, the longevity of the effects makes these preparations suitable for further exploration of the restorative effects of cholinergic treatments.