The applicability of the European Soil Erosion Model (EUROSEM) in Kenya was tested on two subcatchments: Embori, which lies within a large-scale commercial wheat- and barley-growing region on the slopes of Mt Kenya; and Mukogodo, a dry rangeland region used for communal grazing, which lies at lower altitudes. For both catchments, EUROSEM was used to simulate interrill erosion from single storm events on plots of 10 m by 2 m. The model was calibrated and validated for a barley crop at Embori and three treatments of bare, grass and shrub covers at Mukogodo. The results indicated that EUROSEM was applicable to the Embori catchment, as correlation coefficients of 0·91 and 0·84 were obtained between observed and simulated runoff and soil loss, respectively. However, for Mukogodo, the model did not adequately predict soil loss from vegetated plots having grass or shrubs, although it could predict runoff from bare plots and grass quite well (R2 = 0·97, R2 = 0·60, respectively) as well as soil loss from bare plots (R2 = 0·58), but the standard errors of estimate at 4·67 mm were rather high (P = 0·95). Thus, EUROSEM requires improvements in terms of being able to predict runoff and soil loss in semi-arid regions of Kenya, so that it can accommodate multiple vegetation covers, rangeland conditions and soils with surface sealing properties. In addition, the model requires further testing to sample a wider range of climatic and land use types for it to be applicable to Kenya and rangeland conditions generally.