Vegetation cover acts in a complex way in influencing runoff and soil loss and a great deal of information is needed to model these effects. In the Mediterranean, the abandonment of land is important under extensive landuse. Abandoned lands typically have a rolling landscape with steep slopes, and are dominated by herbaceous communities that grow on pasture land interspersed by shrubs. To characterize communities of vegetation such as these, which grow in central Spain, and to evaluate their direct and indirect effects on runoff and soil loss, we carried out experiments with simulated rain. We assessed separately the effects of pasture land and of four species of shrubs (Dorycnium pentaphyllum Scop., Medicago strasseri Greuter et al., Colutea arborescens L. and Retama sphaerocarpa, L.). The infiltration rates under herbaceous vegetation were 7.9 times greater than those obtained on bare land (92.2 mm hour −1 compared with 11.7 mm hour −1 ), and 88% of these differences could be attributed to direct effects. On the pasture land, as the proportion of covered land increased, the runoff decreased linearly, whereas the soil loss decreased exponentially. On the land covered by shrubs, the average infiltration rate was 82.5 mm hour −1 . Under D. pentaphyllum and M. strasseri infiltration rates were greater than 105 mm hour −1 , whereas for R. sphaerocarpa the infiltration rate was 57 mm hour −1 . For D. pentaphyllum and M. strasseri soil loss was less than 4.5 g m −2 , whereas for C. arborescens soil loss was 61.4 g m −2 . Unlike the results for the pasture land, for the shrub-type vegetation the increases in infiltration rates could be attributed to indirect effects: they explained 47% of the increase in infiltration for C. arborescens, 69% for R. sphaerocarpa, 75% for D. pentaphyllum and 100% for M. strasseri.the relationship between vegetation cover and soil loss through an exponential equation (Soil loss = e −2.5 SoilCover/100 ). Despite its simplicity, this is a valuable equation and has been used by Nearing et al. (1989), amongst others, for the development of the WEPP model. However, because it only includes soil cover, corrections are often necessary, because relationships between vegetation and soil loss are complex.Vegetation cover also has indirect effects on soil loss. Root systems and organic matter inputs modify soil porosity, biological activity, aggregate stability and water-holding capacity. In the short term, the influence of vegetation acts mainly through cover effects, but indirect effects become more important with time (Bochet et al., 2006). When the relationship between vegetation and infiltration is analysed, as found with soil loss, an exponential pattern can be observed. Increases in infiltration rates result partly from physical canopy or litter effects, and there is a large influence from soil modifications that appears in the mid-and long-term through, for example, changes in organic matter content or bulk density.