Vegetation restoration and construction play an important role in controlling soil and water loss. The spatial distribution of plant cover is scattered in arid and semi-arid region. The formed land cover patterns on hillslope such as bare soil-vegetation mosaic and banded vegetation pattern have great effect on runoff and soil erosion processes. The relationship between land cover pattern and soil loss is an important topic of pattern and process research in landscape ecology. We present a review of previous research from three aspects including the effects of vegetation patch and land cover pattern on runoff and soil erosion, and the way to account for land cover pattern and couple it with soil loss. Vegetation type, vegetation structure and vegetation morphology are the key factors of soil and water loss at the patch scale. Accurately understanding the runoff and sediment production characteristics of different land cover type and describing land cover pattern are two important bases to study the soil and water loss rules of land cover pattern on hillslope. The land cover pattern changes the continuity of runoff and sediment transport and convergence path. Thus, one should pay much more attention to the role of connectivity and distribution of runoff and sediment source and sink areas on hillslope in affecting soil and water loss. On the base of precise hillslope land cover mapping, developing pattern metrics involved soil loss processes and integrating land cover information in runoff and soil erosion models are two efficient approaches to quantitatively account for the effect of land cover pattern. There are several issues need further investigation in future. First, land cover pattern has a temporal variation which was ignored in most of previous studies. More studies are needed to investigate the influence of land cover pattern variation on soil and water loss from a systemic and dynamic view. Second, current studies on the interaction between land cover pattern and soil loss are mainly focused on the one-way effect analysis, e.g., impact of land cover pattern on soil loss. The feedbacks of soil loss processes to land cover pattern are ignored. Much more attention should be paid to understand the interactions and feedbacks between land cover pattern and soil loss. Third, scale is a universal appearance in ecological phenomenon. Investigating scale effect of the relationship between land cover pattern and soil loss and developing up-scaling methods are front and challenging topics. Fourth, much more simple and practical pattern metrics with physical mechanism should be developed to integrate land cover pattern and soil loss, and pattern metrics should be further applied to quantitatively indicate runoff and sediment yield. Finally, the variation of land cover pattern should be coupled with runoff and sediment production processes in hydrologic model. It can strengthen the ability of the model to account for the feedbacks between pattern and runoff and sediment. This is the key to establish the land cover pattern-soil loss process integrative model, which is very useful to provide scientific basis for protecting soil erosion and designing vegetation rehabilitation pattern in arid and semi-arid region.