The finite spatial resolution of grid-based numerical chemical transport models can lead to overestimation of chemical reaction rates that are in reality limited by the rate of mixing. A parameterization has been developed to compensate for this effect. It estimates the time taken to mix chemical features down from the model grid scale to the physical scale at which they really react, then builds an appropriate time lag into the chemistry by reducing the rate constant. The parameterization is shown to work in idealized tests, where it allows low-resolution parameterized experiments to mimic the results of higher-resolution unparameterized experiments in terms of the total reaction product. The parameterization is applied in a single-layer chemical transport model of the winter stratosphere to the deactivation of ClOx by reaction with NOx. It is found to have a large effect on the deactivation of ClOx. The resulting impact on the rate of catalytic destruction of O3 by ClOx is estimated to be significant.¿ 1997 American Geophysical Union