Solubility measurements for the assemblage talc + quartz were used to investigate the speciation of magnesium in NaCl-MgCl2 solutions at 300, 350, and 400°C, 500 bars, for total chloride concentrations <1 m. At constant temperature and pressure, measured pH (25°C) systematically decreases upon increasing the Mg Cl ratio of the solution, consistent with buffering of aMg2+ a2H+ by talc-quartz equilibria. Measured quartz solubilities increase with increasing concentration of total dissolved salts (NaCl + MgCl2), suggesting a salting-in behavior for aqueous SiO2 at these conditions. A thermodynamic analysis of talc-quartz-solution equilibria in the Na2O-MgO-SiO2-H2O-HCl system indicates that ion-pairing between Mg2+ and Cl- is necessary to account for our measured talc solubilities. At 300, 350, and 400°C, 500 bars, the experimental results are consistent with log K values for MgCl+ dissociation of -2.30, -2.83, and -3.40, respectively. The MgCl02 ion-pair was found to be unnecessary in fitting the data. Theoretical predictions of the dissociation constant for MgCl+ computed from an equation of state for aqueous ions and complexes are in good agreement with these experimental values. Fluid speciation calculations show that MgCl+ predominates over Mg2+ in NaCl-MgCl2 solutions with near neutral pH and chloride concentrations close to those for seawater (~0.5 m), representing 70 to 80% of the total Mg at 300 to 400°C, 500 bars. The retrieval of thermodynamic data for Mg minerals from solubility measurements in chloride solutions must, therefore, take into account MgCl+ explicitly. We apply our results to the stability of caminite (magnesium-hydroxide-sulfate-hydrate), an important constituent of hydrothermal vent deposits on the seafloor. At 300 to 350°C, 500 bars, the stability of caminite is predicted to increase relative to estimates originally computed from solubility measurements in hydrothermal seawater experiments where no provision was made for ion-pairing between magnesium and chloride. The data presented in this study should, therefore, improve the accuracy of mass-transfer calculations designed to predict the fate of seawater-derived Mg in the oceanic crust.