Magnesium transport across sarcoplasmic reticulum (SR) vesicles was investigated in reaction mixtures of various composition using antipyrylazo III or arsenazo I to monitor extravesicular free Mg2+. The half-time of passive Mg2+ efflux from Mg2+-loaded SR was 100 s in 100 mM KCl, 150 S in 100 mM K gluconate, and 370 S in either 100 mM Tris methanesulfonate or 200 mM sucrose solutions. The concentration and time course of Mg2+ released into the medium was also measured during ATP-dependent Ca2+ uptake by SR. In reaction mixtures containing up to 3 mM Mg2+, small changes in free magnesium of 10 microM or less were accurately detected without interference from changes in free Ca2+ of up to 100 microM. Three experimental protocols were used to determine whether the increase of free [Mg2+] in the medium after an addition of ATP was due to Mg2+ dissociated from ATP following ATP hydrolysis or to Mg2+ translocation from inside to outside of the vesicles. 1) In the presence of ATP-regenerating systems which maintained constant ATP to ADP ratios and normal rates of active Ca2+ uptake, the increase of Mg2+ in the medium was negligible. 2) Mg2+ released during ATP-dependent Ca2+ uptake by SR was similar to that observed during ATP hydrolysis catalyzed by apyrase, in the absence of SR. 3) In SR lysed with Triton X-100 such that Ca2+ transport was uncoupled from ATPase activity, the rate and amount of Mg2+ release was greater than that observed during ATP-dependent Ca2+ uptake by intact vesicles. Taken together, the results indicate that passive fluxes of Mg2+ across SR membranes are 10 times faster than those of Ca2+ and that Mg2+ is not counter-transported during active Ca2+ accumulation by SR even in reaction mixtures containing minimal concentrations of membrane permeable ions that could be rapidly exchanged or cotransported with Ca2+ (e.g. K+ or Cl-).