A model is developed to predict protein retention and resolution in mixed-bed columns containing a mixture of a small-pore weak base resin and a large-pore strong base resin. Using amine buffers, the weak base resin generates smooth pH gradients when the column is subjected to pH steps, while the strong base resin interacts with the proteins effecting the separation. The model describes simultaneously ion exchange interaction on the weak base resin and protein adsorption and transport on the strong base resin. The separation of a monoclonal antibody (mAb) from a charge variant whose pI is about 0.2 pH units lower is studied experimentally with the anion exchange resins AG 4-X4 and SOURCE 30Q as the weak and strong base components, respectively. The properties of each resin are characterized independently and model parameters are obtained experimentally for each. Model predictions of pH gradient-based separations in columns containing different proportions of the two resins and at different buffer concentrations are found to be in agreement with experimental results. The model can thus be used as a process design tool to predict combinations of buffer and mixed-bed compositions, column lengths, and mobile phase velocities that would yield a desired separation performance.