Rats received bilateral lesions of the nucleus basalis magnocellularis by infusion of ibotenic acid. Fourteen days later, osmotic minipumps releasing human recombinant nerve growth factor (0.3 micrograms/day) were implanted subcutaneously. Starting one month after the lesion, spatial learning of the animals was tested using the Morris water maze. Acquisition of the task was impaired by the lesion, but treatment with nerve growth factor reduced the average latency to find the platform by approximately 9 s, which represents 28% of the lesion-induced behavioral deficit. Retention of this task and spatial acuity, tested in a trial in which the platform was not present, did not show a statistically significant improvement. Lesions of the nucleus basalis magnocellularis reduced the choline acetyltransferase activity in the neocortex, but not in the hippocampus. Treatment with nerve growth factor increased the choline acetyltransferase activity in the neocortex but not in the hippocampus. There was no significant difference in the levels of norepinephrine, dopamine, serotonin or their metabolites in the cortex or hippocampus between nerve growth factor-treated animals and lesioned control animals. There was no significant correlation between any of these neurochemical changes and behavioral performance (acquisition and spatial acuity). Treatment with nerve growth factor did not increase the number or the size of nerve growth factor receptor-immunoreactive neurons in the nucleus basalis magnocellularis. These data suggest that delayed treatment with nerve growth factor results in an improvement of spatial learning in rats with lesions of the nucleus basalis magnocellularis. A possible role for cholinergic mechanisms in this effect is discussed.