Soil evaporation (E) and plant transpiration (T) dominate the water budget of semi-arid sparse steppe, and they are crucial for vegetation restoration, as shadows cast by sparse vegetation may shade the sunlight and thereby reduce near-ground radiation – which could affect biotic process of plants and land-atmosphere flux transport dominated by E and evapotranspiration (ET). Currently, there are few studies on the effects of shadow on E and ET in semi-arid sparse steppe because of the lack of appropriate research tools. However, high-resolution thermal cameras provide an opportunity to meet this challenge. This study aims to detect the shadow-affected soil area in sparse steppe and determine its effects on E and ET by using the thermal imaging with classification algorithm and the extended three-temperature model (3T model) with the introduction of a new calibration coefficient, which considers the influence of shadows on E and ET. A series of field experiments were carried out in a semi-arid restored grassland in Taipus Banner, Inner Mongolia, China, in summer 2008. The Bowen ratio method was used to validate the ET estimates. Results showed that the importance of shading effects on E and ET tends to increase along with the plant coverage as the restoration prolongs. And shadow-affected soil zone around the plant could be larger than plant area. And ET and E may be 4–28% and 10–66% overestimated if the effect of shadow is not considered. Implication of this study on other ET estimation methods in remote sensing is further discussed. Copyright © 2014 John Wiley & Sons, Ltd.