Effects of a pre-training intraperitoneal glucose injection on learning and memory were tested using two tasks: passive avoidance and Morris water maze. In the former task, mice that had received glucose 2 h prior (but not 1, 3, or 5 h prior) to a trial that combined acquisition with passive avoidance of foot shock showed a significantly increased retention latency when tested 24 h later. Thus, this effect was time-dependent, and it was also found to be dose-dependent by further experiment. In contrast, 2-deoxy-D-glucose and fructose had no such effect. In the Morris water maze task, glucose injection 2 or 3 h before a block of trials enhanced the spatial memory performance of mice. These glucose-induced memory-facilitation effects were abolished by an intracerebroventricular injection of anti-acidic fibroblast growth factor antibody 30 min before the glucose injection, suggesting a critical role for endogenous acidic fibroblast growth factor in this facilitatory effect. Furthermore, continuous intracerebroventricular infusion of acidic fibroblast growth factor in rats significantly increased retention latency (when tested repeatedly on successive days using a passive avoidance task). Our earlier studies demonstrated that brain acidic fibroblast growth factor is produced in the ependymal cells of the cerebroventricular system, and is released into the cerebrospinal fluid following either a meal or a (intraperitoneal or intracerebroventricular) glucose injection. This released acidic fibroblast growth factor also diffuses into the brain parenchyma, and is taken up by neurons in the hippocampus, hypothalamus, and elsewhere in the brain some 2 h after the meal or glucose injection. These and the present findings indicate (i) that pre-training glucose injection improves memory performance, and (ii) that acidic fibroblast growth factor, especially by its action within the hippocampus, is involved in this enhancement process.