With passive avoidance (PA), Morris water maze (WM) and eight-arm radial maze tasks, we evaluated the memory-enhancing action of FK960 [N-(4-acetyl-1-piperazinyl)-p-fluorobenzamide monohydrate], a compound which we have found through rational drug screening based on our hypothesis that penile erection is a valid predictor of central cholinergic activation. Memory performance in the tasks was impaired in aged (24- to 26-months-old) rats as well as in rats with nucleus basalis magnocellularis lesions. Scopolamine (1 mg/kg i.p.) treatment induced memory impairment in PA and WM; treatment with cysteamine (200 mg/kg s.c.) induced memory impairment in PA but not in WM, whereas fimbria fornix lesioning affected the rats in the opposite manner. FK960 (0.1-10 mg/kg i.p.) ameliorated all the memory impairments except those induced by cysteamine or fimbria fornix lesion, and the dose-response curves were bell shaped with maximal response at 1 to 3.2 mg/kg. The effects of FK960 on the scopolamine-induced memory impairment in the PA and/or WM were abolished by cysteamine (200 mg/kg s.c.), dl-p-chlorophenylalanine methyl ester hydrochloride (150 mg/kg i.p. for 3 days) or raphe lesioning, but not by neonatal 6-hydroxydopamine (35 micrograms/head) treatment. Neurochemical analysis revealed that cysteamine and raphe lesions reduced brain somatostatin and serotonin contents, respectively. The treatment with FK960 (0.32-320 mg/kg p.o.) dose-dependently increased both serotonin and 5-hydroxyindoleacetic acid levels in the brain areas examined and significantly increased hippocampal somatostatin contents at the smaller doses. From these results, we conclude that FK960 ameliorates cognitive dysfunction through an activation of the somatostatinergic-serotonergic link.