Soil interrill erodibility is a key component of soil erosion models. However, when using aggregate stability to assess soil erodibility, samples are usually collected from the plough layer, while soil erosion occurs at the soil surface. Hence, the potential changes in erodibility caused by crusting are ignored. Moreover, soil interrill erodibility is difficult to predict accurately. This lack of predictability means that current erosion models use a constant erodibility value for a given soil, and thus do not consider potential heterogeneity of erodibility. This study was conducted to (i) assess the heterogeneity of aggregate stability for a crusted soil and (ii) relate this heterogeneity to the aggregate stability of the underlying material (sub-crust) and to standard soil properties. A field study was conducted in a small area of the Loess Plateau in China in which the crust and the sub-crust soils were sampled. Standard soil properties (organic matter content, sand content, silt content, clay content, cation exchange capacity (CEC), pH in water, and water content at the time of sampling) were measured as potential explanatory factors of aggregate stability. The results showed a large heterogeneity in aggregate stability among the sites, even though the sites had the same soil type. The mean weight diameter (MWD) of the crust varied between 0.33 and 2.04 mm while the MWD of the sub-crust varied between 0.23 and 1.42 mm. Soil texture and pH were uniform among the sampling sites, whereas water content, organic matter content and CEC varied more. Even though some correlations existed (for example r = 0.57 between MWD for the slow wetting test and organic matter content), none of the standard soil properties was able to predict aggregate stability accurately. The aggregate stability of the crust was significantly greater than that of the sub-crust. The large differences in aggregate stability imply large differences in soil interrill erodibility. Because a single soil type was investigated, this finding proves that erodibility can vary greatly in space even for a given soil type. Soil interrill erodibility should be estimated from the exact material exposed to erosive forces, the soil surface material. Using the sub-crust would have led to greatly over-estimated erodibility and thus to a marked bias in erosion model predictions.