Small‐scale aerial photographs and high‐resolution satellite images, available for Ethiopia since the second half of the twentieth century as for most countries, allow only the length of gullies to be determined. Understanding the development of gully volumes therefore requires that empirical relations between gully volume (V) and length (L) are established in the field. So far, such V–L relations have been proposed for a limited number of gullies/environments and were especially developed for ephemeral gullies. In this study, V–L relations were established for permanent gullies in northern Ethiopia, having a total length of 152 km. In order to take the regional variability in environmental characteristics into account, factors that control gully cross‐sectional morphology were studied from 811 cross‐sections. This indicated that the lithology and the presence of check dams or low‐active channels were the most important controls of gully cross‐sectional shape and size. Cross‐sectional size could be fairly well predicted by their drainage area. The V–L relation for the complete dataset was V = 0 · 562 L ^1·381 (n = 33, r² = 0 · 94, with 34 · 9% of the network having check dams and/or being low‐active). Producing such relations for the different lithologies and percentages of the gully network having check dams and/or being low‐active allows historical gully development from historical remote sensing data to be assessed. In addition, gully volume was also related to its catchments area (A) and catchment slope gradient (S_c). This study demonstrates that V–L and V–A × S_c relations can be very suitable for planners to assess gully volume, but that the establishment of such relations is necessarily region‐specific. Copyright © 2013 John Wiley & Sons, Ltd.