Most regional‐scale soil erosion models are spatially lumped and hence have limited application to practical problems such as the evaluation of the spatial variability of soil erosion and sediment delivery within a catchment. Therefore, the objectives of this study were as follows: (i) to calibrate and assess the performance of a spatially distributed WATEM/SEDEM model in predicting absolute sediment yield and specific sediment yield from 12 catchments in Tigray (Ethiopia) by using two different sediment transport capacity equations (original and modified) and (ii) to assess the performance of WATEM/SEDEM for the identification of critical sediment source areas needed for targeting catchment management. The performance of the two model versions for sediment yield was found promising for the 12 catchments. For both versions, model performance for the nine catchments with limited gully erosion was clearly better than the performance obtained when including the three catchments with significant gully erosion. Moreover, there is no significant difference (alpha 5 per cent) between the performances of the two model versions. Cultivated lands were found to be on average five times more prone to erosion than bush shrub lands. The predicted soil loss values in most parts of Gindae catchment are generally high as compared with the soil formation rates. This emphasises the importance of implementing appropriate soil and water conservation measures in critical sediment source areas prioritising the steepest part of the catchment (i.e. areas with slope >50 per cent). The applicability of the WATEM/SEDEM model to environments where gully erosion is important requires the incorporation of permanent gully and bank gully erosion in the model structure. Copyright # 2011 John Wiley & Sons, Ltd.