The majority of neurotransmitters mediating fast neurotransmission in the central nervous system are amino acids (glutamate, GABA and glycine). Their metabolism and exchange between cells suggest there might be high fluxes of ammonium (NH₄⁺/NH₃) in the brain. The recent reports of membrane transporters which can transport ammonium selectively have increased interest in the role of ammonium as a potential intercellular messenger in normal brain function. An intercellular flux of ammonium has been described in the bee retina. In this tissue, exogenous glucose is not significantly taken up by neurons (the photoreceptors), which rely on adjacent glial cells for supply of an amino acid, alanine, as an energetic substrate. When photoreceptors are activated, ammonium is released, presumably as a result of the metabolic oxidation of alanine. Ammonium is then avidly taken up into glial cells through a selective route, a NH₄-Cl cotransporter. This ammonium flux has been recently monitored in real time with ammonium-sensitive microelectrodes in retinal slices, showing that it is controlled by the glial NH₄-Cl cotransporter. In mammals, about 90% of central synapses are glutamatergic. Following glutamate release into the extracellular space, glutamate is taken up through transporters, mainly into glial cells (Marcaggi & Attwell, 2004). In order to recycle glutamate to the nerve terminals, a flux of glutamine in the opposite direction has been suggested because of the preferential localization of glutamine synthetase (converting glutamate and ammonium into glutamine) in glial cells and glutaminase (converting glutamine into glutamate and ammonium) in neurons (Marcaggi & Coles, 2001). This recycling, called glutamate-glutamine cycle, implies a production of ammonium by neurons and use of ammonium by glial cells, and thus requires a nitrogen flux from neurons to glial cells. Alanine has recently been proposed as the nitrogen carrier mediating this nitrogen flux. However, this proposal is not supported by the recent characterization and localization of putative alanine transporters (Marcaggi & Attwell, 2004). I therefore hypothesize that ammonium is mediating the nitrogen flux from neurons to glial cells. This flux might be quantitatively as important as the flux of synaptic glutamate release, which was estimated to be as high as the flux of brain glucose consumption (Sibson et al. 1998). To date, ammonium release as a result of neuronal activity has only been shown in the frog sciatic nerve and the bee retina. However, it can be predicted that ammonium will be released during synaptic activity for the following two reasons. First, the production of presynaptic glutamate and GABA from precursor glutamine will produce ammonium in glutamatergic and GABAergic terminals. Second, exogenous NH₄Cl neutralizes the acid pH of synaptic vesicles (Li et al., 2005), implying a vesicular membrane permeability to NH₃. [NH₃] in vesicles can therefore be expected to equilibrate with [NH₃] in the surrounding cytoplasm, and a NH₄⁺ gradient will be established due to the pH difference (ΔpH) between the cytoplasm and the lumen of the vesicles (typically ~ 1.5), with [NH₄⁺]vesicle/[NH₄⁺]cytoplasm = 10^ΔpH. Thus, ammonium will be concentrated in vesicles and might be expected to be released during synaptic transmission. Cultured astrocytes have been reported to transport NH₄⁺ through non-selective routes, driven by the membrane potential and Cl^– gradient (Marcaggi & Coles, 2001). Although this remains to be confirmed for glial cells in intact nervous tissue, such transport would support a flux of ammonium from neurons to glial cells as described in the bee retina. Ammonium can activate three enzymes of glycolysis: hexokinase, phosphofructokinase, and pyruvate kinase (Muntz & Hurwitz, 1951). Ammonium is thus a good candidate messenger to modulate glial metabolism locally in response to synaptic transmission and neuronal activity.