The Oak Ridge National Laboratory high-temperature isopiestic apparatus was outfitted with precise pressure gauges to allow for direct vapor pressure measurements. Vapor pressures over concentrated solutions of CaCl2(aq), and CaBr2(aq) were measured at temperatures between (380.15 and 523.15) K in the range of water activities between 0.2 and 0.85. Isopiestic molalities were used to determine osmotic coefficients at the conditions where NaCl reference standard solutions remained undersaturated. The main goal of this work was to improve the accuracy of isopiestic comparisons based on the calcium chloride reference standard. Osmotic coefficients for CaCl2(aq) and CaBr2(aq) calculated from both isopiestic and direct vapor pressure results were combined with the literature data and used to build general thermodynamic models based on a variant of extended Pitzer ion-interaction equations and valid at the saturation pressure of water. While these empirical models approach the accuracy of the experimental data in a wider range of concentrations and temperatures than any previously published equations, considerable amounts of accurate data and a substantial effort are required in order to obtain a satisfactory representation using power series-based virial equations. The effect of experimental uncertainties on the accuracy of the direct vapor pressure results is discussed, including in particular the error caused by the presence in the apparatus of a small amount of CO2. The substantial decrease of the solubility product of CaCO3 in concentrated chloride solutions at temperatures above 423 K is a serious defect of calcium chloride as a water activity reference standard. Published by Elsevier Ltd.