In this paper the numerical model presented in the companion paper is tested and applied. Assessment of model accuracy was based on two approaches. First, predictions of evolution of a 13.5 km reach of the South Fork of the Forked Deer River, in west Tennessee, were compared to observations over a 24-yr period. Results suggest that although the model was able to qualitatively predict trends of widening and deepening, quantitative predictions were not reliable. Simulated widths and depths were within 15% of the corresponding observed values, but observed change in these parameters at the study sites were also close to these values. Simulated rates of depth adjustment were within 15% of observed rates, but observed rates of channel widening at the study sites were approximately three times those simulated by the model. In the second approach, the model was used to generate relationships between stable channel width and bank-full discharge. The model was able to successfully replicate the form of empirically derived regime-width equations. Simulations were used to demonstrate the model's ability to obtain more realistic predictions of bed evolution in widening channels.