Dry water repellent soils are known to inhibit water in®ltration, ultimately forcing water to¯ow via preferential paths through the vadose zone. To study water¯ow and transport in a water repellent sandy soil, a bromide tracer experiment had been carried out, which started in the fall after winter wheat had been sown. Despite the uniform tracer application, soil core sampling indicated that bromide concentrations varied largely from place to place. Wetter sites in the experimental ®eld received more bromide, due to lateral transport through a thin top layer. Wetting fronts in®ltrated deeper here, leading to perturbed wetting fronts in the experimental ®eld. In contrast to what was expected, the wetting front perturbations did not grow to ®ngers. Numerical results indicate that this was attributed to the relatively high soil water contents during the experiment, which caused the soil to be wettable instead of water repellent. The water-entry capillary pressure of the secondary wetting branch exceeds the air-entry capillary pressure of the primary drainage branch in this case. In the opposite situation, with the water-entry capillary pressure of the secondary wetting branch beneath the air-entry capillary pressure of the primary drainage branch, perturbations would have grown to ®ngers. Such a situation occurs during in®ltration in initially dry, water repellent soil. The results presented illustrate the effect of antecedent moisture conditions on the formation of stable and unstable wetting fronts, and its relation to the moment of tracer application. #