The involvement of glutamate mediated neurotoxicity in the pathogenesis of Alzheimer's disease is finding increasingly more acceptance in the scientific community. Central to this hypothesis is the assumption that in particular glutamate receptors of the N-methyl-D-aspartate (NMDA) type are overactivated in a tonic rather than a phasic manner. Such continuous mild activation leads under chronic conditions to neuronal damage. Moreover, one should consider that impairment of plasticity (learning) may result not only from neuronal damage per se but also from continuous activation of NMDA receptors. To investigate this possibility we tested whether overactivation of NMDA receptors using either non-toxic doses/concentrations of a direct NMDA agonist or through an indirect approach--decrease in magnesium concentration--produces deficits in plasticity. In fact NMDA both in vivo (passive avoidance test) and in vitro (LTP in CA1 region) impaired learning and synaptic plasticity. Under these conditions memantine which is an uncompetitive NMDA receptor antagonist with features of "improved magnesium" (voltage dependence, affinity) attenuated the deficit. The more direct proof that memantine can act as a surrogate for magnesium was obtained in LTP experiments under low magnesium conditions. In this case as well, impaired LTP was restored in the presence of therapeutically relevant concentrations of memantine (1 microM). In vivo, doses leading to similar brain/serum levels produce neuroprotection in animal models relevant for neurodegeneration in Alzheimer's disease such as neurotoxicity produced by inflammation in the NBM or beta-amyloid injection to the hippocampus. Hence, we postulate that if in Alzheimer's disease overactivation of NMDA receptors occurs indeed, memantine would be expected to improve both symptoms (cognition) and slow down disease progression because it takes over the physiological function of magnesium.