Land degradation is difficult to assess in arid rangelands because of short-term variations in rainfall, landscape diversity and the problems of sampling very large areas. This paper shows how vegetation cover index values derived from multi-temporal remotely-sensed data can be used in association with spatial models of grazing impact on landscapes to identify grazing-induced land degradation. The method assumes that grazing effects decrease with distance from water and that temporary grazing impacts largely disappear as vegetation responds to major rainfalls. Grazing gradients (i.e., systematic changes in vegetation cover with distance from water) which remain indicate long term damage. Simple grazing gradients involve changes in average cover with distance from water and include normal, inverse and composite types, complex grazing gradients show systematic changes in cover variance with distance from water and develop where soil and runoff are being redistributed. Although complex grazing gradients may involve little change in total cover with distance from water, they are symptomatic of a reduction in the proportion of forage present. The ability to recognize grazing gradients may greatly simplify range assessment procedures and may also improve satellite image based procedures for determining erosion risk.