Most soil-erosion research has been undertaken at small spatial scales, i.e. from small erosion plots over field parcels to small catchments of a few hectares. On the other hand, sediment-yield data for several large river basins (>1000 km2) are also available. Little information exists, however, on the link between soil-erosion processes on hillslopes and yields of suspended sediment in large rivers. Data on sediment yield for smaller catchments (1–50 km2) are therefore needed. Such data can be obtained by different measuring methods of which sediment rating curves or river sampling are the most common. In this study, however, the possibilities and limitations of the use of sedimentary deposits in small ponds (<1 ha) to assess sediment yields is investigated. In small catchments (<50 km2), many small flood retention ponds, ponds for irrigation purposes or farm ponds already exist and no expensive monitoring equipment needs to be installed to assess sediment yield. This technique, however, also has a few limitations. Volumetric sediment deposition rates measured in these ponds need to be converted to sediment masses using representative values of the dry sediment bulk density. Furthermore, measured sediment masses need to be corrected for the trap efficiency of the pond otherwise sediment yield will be underestimated. These conversions and corrections are subject to errors. For 21 catchments in central Belgium, the mean error on computed sediment yield, using sediment deposits in small ponds, was estimated to range between 40 and 50 per cent. This error more or less corresponds to the error when using sediment rating curves and suggests that this technique can be a valuable alternative to provide data on the regional variations in sediment yield. Copyright © 2002 John Wiley & Sons, Ltd.