Experiments were designed to determine whether the putative Na(+)-Mg2+ exchanger previously demonstrated to mediate Mg2+ efflux (R. DiPolo and L. Beagué. Biochim. Biophys. Acta 946: 424-428, 1988) could also mediate the efflux of Na+ (presumably a Na+ efflux-Mg2+ influx exchange) in squid giant axons. The effects of external Mg2+ (Mg(o)) on 22Na efflux were measured in internally dialyzed, ATP-fueled axons in which the contribution to Na+ efflux by other pathways was inhibited. To facilitate measurement of Mg(o)-dependent Na+ efflux, the intracellular concentration of Na+ was increased. To prevent Na(+)-Na+ exchange, external Na+ was replaced by tris(hydroxymethyl)aminomethane. To assess the effect of Mg(o) on Na+ efflux without altering the total divalent cation concentrations, Mg(o) was replaced mole-for-mole by external Ba2+ (Ba(o)). This manipulation produced reversible reductions in Na+ efflux. These reductions were neither due to membrane hyperpolarization nor to a direct effect of Bao but were due instead to the reduction in Mg(o). The Mg(o)-dependent Na+ efflux was inhibited by external amiloride but was spared by bumetanide. In the absence of external Na+, the Mgo-dependent Na+ efflux increased as a function of external Mg2+ with Michaelis-Menten kinetics. These results indicate that the Na(+)-Mg2+ exchange can mediate the efflux of Na+ (operate in Na+ efflux-Mg2+ influx mode of exchange).