Based on bedform experiments in a large-scale flume, we demonstrate that the rate of development of wave ripples on a mixed sand–clay bed under regular waves is significantly lower than on a pure-sand bed, even at clay fractions as low as 4.2%, and that this rate of development decreases exponentially from 4.2% to 7.4% clay. These experiments also showed that, despite the slow growth of the bedforms in the mixed sand–clay, the equilibrium length and height of the wave ripples were independent of the initial bed clay fraction. Given that the ripple crests were composed of pure sand at the end of all the experiments that started with well-mixed sand–clay, it is inferred that the clay was removed from the bed during the development of the wave ripples through winnowing into the water column, and possibly also by sieving into the subsurface, where the final clay fractions were found to be higher than the initial clay fractions. These clay removal processes are interpreted to have facilitated the wave ripples to reach equilibrium lengths and heights that are similar to those in pure sand. Clay- carrying pore flow initiated by pressure gradients between the wave ripple troughs and crests might also have contributed to the accumulation of clay in the sediment below the wave ripples. The formation of the clay-enriched ‘armoring’ layer in the substrate is likely to further reduce erosion rates and could influence the dispersion of nutrients and pollutants in coastal seas.