The model ratios applicable to dispersion studies in distorted hydraulic models are investigated for both steady and oscillatory flow, and the effects of both vertical and transverse distributions of velocity on the dispersion process are considered. It is shown that simple transposition of concentration profiles from model to prototype is not valid and that models may magnify or reduce longitudinal dispersion depending on the geometry of the system. For that reason, physical models are unlikely to be useful for studies of pollutant dispersion.