Neuroanatomical methods were used to determine if cocaine irreversibly injures neurons. Despite acute and chronic high-dose treatments for months that produced stereotyped behavior and seizures, and the use of a sensitive silver impregnation method, we were unable to find any evidence of neuronal damage anywhere in the brain. Since expression of the inducible 72 kDa heat shock protein (HSP72) is a sensitive indicator of potentially toxic neuronal stress, we next determined if cocaine evoked HSP72 expression. Even high doses of cocaine that evoked seizures did not induce HSP72 immunoreactivity anywhere within the brain, whereas kainic acid produced widespread HSP72 immunoreactivity and irreversible injury. Having failed to find indications of frank neurotoxicity, we examined peptide and protein cell marker immunoreactivities in search of cocaine-induced changes. Although cocaine treatment had no obvious effects on the patterns of hippocampal calbindin-D28K, somatostatin-, tyrosine hydroxylase- and parvalbumin immunoreactivities, cocaine reliably altered neuropeptide Y-like immunoreactivity (NPY-LI). Most notably, NPY-LI was expressed in hippocampal dentate granule cells and pyriform cortical neurons, which do not normally express it. Conversely, we noted decreased NPY-LI in dentate hilar neurons that normally do express it. Since both changes in NPY-LI were seen only in cocaine-treated rats that exhibited seizures, the role of seizure activity per se in producing the NPY changes was addressed in normal rats by electrical stimulation of the perforant path. Like cocaine, perforant path stimulation for as little as 15min evoked NPY-LI in granule cells but did not replicate the cocaine-induced decrease in hilar cell NPY-LI. These results suggest that cocaine does not irreversibly injure neurons in the rat, even at doses that induce seizures. However, cocaine produces long-lasting changes in NPY expression that are of unknown functional significance. Our inability to demonstrate cocaine-induced neuronal damage in rats should in no way be taken as evidence of its safety in humans.