Balancing erosion control and the water conservation service is the primary goal of intensive revegetation in water-limited conditions. In this study, a field experiment was conducted in the dryland Chinese Loess Plateau (CLP), to investigate the role of vegetation pattern and density in their tradeoff between soil conservation (SC) and soil water retention (SWR). Five vegetation patterns (up-slope covered, downslope covered, low and high uniformly covered, no coverage) with three levels of grass density (220 g m À2 , 110 g m À2 , and 0) were designed and implemented. A novel indicator EEW (the exchange between erosion reduction and water consumption), as the cost of any extra depletion of soil water to improve each unit of soil conservation, was used to represent the tradeoff during a complete dry-wet-dry soil water convert cycle. The higher EEW, the poorer tradeoff between SC and SWR. It was illustrated that higher vegetation density could promote SC but not inevitably suppress EEW.The present typical grass density in CLP, round 220 g m À2 , was redundant in optimizing the SC-SWR tradeoff. A concentrated and centralized vegetation distribution mainly in the downslope obtained an EEW at 11.6 mm m 2 kg À1 and was confirmed as a favoured vegetation pattern. The simulated typical density and spatial pattern in CLP (about 220 g m À2 and uniformly distribution) had EEW at 20.5 mm m 2 kg À1 , by reducing the density to about 110 g m À2 and a downslope centralized vegetation redistribution, the EEW was reduced by 43.4%. Under long-term drought conditions at dry-wet-dry cycles, soil water decreased more sharply in the Full-covered plots, due to more evapotranspiration and vertical water loss caused by higher plant density, compared with half-covered and bare soil plots. The present study confirmed that planted density and vegetation pattern can manipulate SC-SWR tradeoff and it was thus possible to optimize it by adjusting revegetation's intensity and structures in practice.