1. Mg2+ efflux from ferret red cells into a nominally Mg2(+)-free medium is 41 +/- 2 mumol (l cell)-1 h-1. The properties of Mg2+ transport can be measured in these cells without the need for Mg2+ loading. 2. Amiloride, quinidine, imipramine and external divalent cations partially inhibit Mg2+ efflux. Maximal inhibition by these agents is about 60-70% suggesting that at least two Mg2+ transport pathways exist. 3. As external Na+ is replaced by choline or N-methyl-D-glucamine Mg2+ efflux is first stimulated, reaching a peak when external [Na+] ([Na+]o) is about 10 mM, and then inhibited. Mg2+ transport reverses direction so net Mg2+ uptake occurs when [Na+]o is reduced below 1 mM. 4. Mg2+ efflux is stimulated when 0.1 mM-EDTA is added to the medium only when [Na+]o is low. 5. Reduction of cell ATP content to about 20 mumol (l cell)-1 by treating cells with 2-deoxyglucose stimulates Mg2+ efflux measured over the 2 h period following depletion. 6. Substantial Mg2+ influx can be observed in ferret red cells when they are incubated in media containing 10 mM-Mg2+. Influx is stimulated by reducing [Na+]o to 10 mM. Further reduction of [Na+]o to below 1 mM reduces Mg2+ uptake. A component of uptake is inhibited by external Co2+. 7. Na(+)-Mg2+ antiport may account for a substantial component of Mg2+ transport in ferret red cells. The direction of transport can be reversed by sufficiently lowering [Na+]o or by increasing external [Mg2+]. Analysis of the conditions at which transport reverses direction suggests transport with a stoichiometry of 1 Na+:1 Mg2+. Antiport with this stoichiometry would also explain maintenance of the physiological level of intracellular ionized Mg2+ in these cells.