The potential for using chlorofluorocarbons (CFCs) to assess ocean climate models is explored within a series of global ocean experiments. The ocean model simulations are run under identical wind stress and thermohaline forcing, but with three different formulations of subgrid-scale mixing. Two passive tracers are included in the equilibrated models, representing the dissolved concentrations of CFC-11 and CFC-12 in seawater. The resulting model simulations are then compared directly with observed CFC levels in key areas for deep and bottom water formation. CFC-11 uptake is found to be overestimated in the Southern Ocean when mixing rates are defined in traditional Cartesian co-ordinates. The inclusion of an isopycnal mixing scheme (which is often used in coupled ocean-atmosphere models) actually degrades the CFC-11 simulation by blending water masses too strongly, particularly in the Southern Ocean. The spurious uptake of CFC-11 at 55¿--70 ¿S suggests that certain climate models might overestimate the role of the Southern Ocean in moderating climate change. A more sophisticated mixing parametrization that simulates the effects of subgrid-scale eddies on the mean ocean flow (and allows for zero lateral diffusion) is seen to greatly reduce CFC-11 uptake in the Southern Ocean. Climate models that adopt this new mixing scheme are likely to predict a more rapid CO2-induced warming over the Southern Hemisphere. ¿ American Geophysical Union 1995