The inverse texture effect of soil on vegetation has been reported in a few arid regions, but no information is available on the vast Chinese temperate grassland. In addition, no consistent criteria and approaches have been used to classify soil texture groups, determine the precipitation threshold (Pθ) below which the inverse texture effect occurs, and quantify the effect size (Z), which has limited the generalization of the rules of the soil texture effect in different ecosystems. We analyzed the soil texture effect on vegetation using the remotely sensed vegetation data (MODIS/NDVI) and the Chinese soil profile texture data at 6851 sites in temperate grasslands of China. We found, first, a strong inverse texture effect, i.e., a higher NDVI on coarse- (>50 % sand) than fine-textured soil in low precipitation areas, and Pθ was variable with annual precipitation and averaged 260 mm in the region. Second, we proposed a novel method to use the loam soil as a benchmark against which to quantify the effect of various textured soils, and determine Pθ as the point on precipitation gradient where the soil texture effect shift from positive to negative or vice versa. As such, Pθ decreased with soil coarseness, and increased with precipitation; and the texture effect decreased from positive to negative for sandy soils, whereas it increased from negative to positive for silt or clay soils, on the gradient of precipitation increase. Our results provide new insights into the relations of vegetation pattern and dynamics with precipitation and soil texture, and a method for quantifying and comparing soil texture effect across ecosystem types.