Analysis of fluid inclusions from porphyry copper deposits reveals that magmatic vapor and brine are vital for the removal of copper from arc magmas and its transport to the site of ore deposition. Experiments in melt-vapor-brine systems allow for investigation of the partitioning of copper between silicate melts and volatile phases at magmatic conditions. The presence of CO2 affects both the pressure at which a melt saturates with respect to volatile phases. Therefore, the partitioning of copper among felsic (rhyolitic) melt, vapor and brine in CO2-bearing experiments was examined to provide insights into copper partitioning and the generation of porphyry copper and related deposits. We present results from experiments performed at 800 degrees C and 100 MPa in CO2-bearing melt-vapor-brine systems with X-CO2(v+b) = 0.10 and 0.38. The compositions of vapor and brine inclusions, and run-product glasses, were determined during the course of this investigation. Microthermometric measurements of fluid inclusions show an increase in the salinity of the magmatic brine (similar to 65 to similar to 70 wt% NaCleq) and decrease in the salinity of the vapor (similar to 3.5 to similar to 1 wt% NaCleq) as X-CO2 is increased from 0.10 to 0.38. The partitioning of copper between brine and vapor (D-Cu(b/v) +/- 2 sigma) increases from 25 (+/- 6) at X-CO2 = 0.10, to 100 (+/- 30) at X-CO2 = 0.38. The partitioning of copper between vapor and melt (D-Cu(v/m) +/- 2 sigma) decreases from 9.6 (+/- 3.3) at X-CO2 = 0.10, to 2 (+/- 0.8) at X-CO2 = 0.38. These data demonstrate that copper partitioning in sulfur-free, CO2-bearing systems is controlled by the changes in the salinity of the vapor and brine that, in turn, are functions of X-CO2. No change in the apparent equilibrium constants for Cu-Na exchange was observed in Fe-bearing experiments which supports a salinity-dependent model for copper partitioning. An existing model (MVPart) for ore metal partitioning between melt and volatile phases was modified to incorporate partitioning data from CO2-bearing experiments. The model can be used to predict trends in efficiency of removal of copper from melts into exsolving CO2-bearing magmatic volatile phases at a fixed pressure and temperature. The CO2-MVPart model predicts that CO2-rich (X-CO2 = 0: 38) magmatic vapor will remove half of the available copper from the melt, without a contribution from the brine, compared to low CO2 (X-CO2 <= 0.10) systems. Thus, periods of CO2-rich vapor exsolution are not expected to be associated with efficient removal of copper from shallow felsic melts. Published by Elsevier Ltd.