Colchicine has been shown to be neurotoxic to cholinergic neurons in the medial septum 1 week following intracerebroventricular injections. The experiments described here were designed to examine the selectivity of this effect over a longer time course, and to examine the role of axoplasmic transport in the neurotoxic effect. As previously reported, 1 week after intracerebroventricular injections of colchicine, the numbers of choline acetyltransferase (ChAT)-immunoreactive neurons in the medial septum-diagonal band complex (MSDB) were reduced to 38% of control; this reduction was stable 2 and 3 weeks post injection. Injections of colchicine placed into the body of the fornix produced similar results. GAD-immunoreactive somata, the other major population of neurons in the MSDB, were unaffected 3 weeks following colchicine, as previously reported 1 week following similar injections. The normal AChE staining pattern in the hippocampus, particularly the dentate gyrus, was depleted following either ICV or intrafornical injections of colchicine. This depletion was more severe with longer survival times. Injections of lumicolchicine, an isomer of colchicine which does not bind tubulin, had no effect on ChAT-immunoreactive neurons in the MSDB or on AChE staining in the hippocampus. Injections of colchicine, but not of lumicolchicine, partially blocked the retrograde transport of the fluorescent dye Fluoro-Gold from the hippocampus to the MSDB. In addition, the content of NGF in the hippocampus rose 84% above control values 2 weeks following colchicine and remained elevated at three weeks. Together these results indicate that colchicine is selectively toxic for cholinergic neurons in the septohippocampal system, and suggest that the alkaloid's neurotoxic effects work via the blockade of axoplasmic transport.