Rainfall experiments have been conducted in the laboratory in order to assess the hydrological response of top soils very susceptible to surface sealing and containing rock fragments in different positions with respect to the soil surface. For a given cover level, rock fragment position in the top soil has an ambivalent effect on water intake and runoff generation. Compared to a bare soil surface rock fragments increase water intake rates as well as time of runoff concentration and decrease runoff volume if they rest on the soil surface. For the same cover level, rock fragments reduce infiltration rate and enhance runoff generation if they are well embedded in the top layer. The effects of rock fragment position on infiltration rate and runoff generation are proportional to cover percentage. Micromorphological analysis and measurements of the saturated hydraulic conductivity of bare top soils and of the top layer underneath rock fragments resting on the soil surface reveal significant differences supporting the mechanism proposed by Poesen (1986): i.e. runoff generated as rock flow or as Horton overland flow can (partly) infiltrate into the unsealed soil surface under the rock fragments, provided that they are not completely embedded in the top layer. Hence, rock fragment position, beside other rock fragment properties, should be taken into account when assessing the hydrological response of soils susceptible to surface sealing and containing rock fragments in their surface layers. A simple model, based on the proportions of bare soil surface, soil surface occupied by embedded rock fragments, and soil surface covered with rock fragments resting on the soil surface, describes the runoff coefficient data relatively well.