Soil erosion is a key concern with regard to ecosystem functionality and food, fibre and bioenergy productions worldwide. Therefore, understanding the mechanisms and controls of soil erosion, particularly the link between soil aggregate stability and soil erodibility, is of utmost importance. The use of disturbed samples and sieved soil to determine the involved erodibility and aggregate stability is standard in soil erosion studies. However, soil erodibility estimation based on disturbed‐soil samples can be inaccurate as it involves changes in the architecture of the considered soil, possibly leading to overestimations. Moreover, a necessity for evaluating soil erodibility beyond intrinsic soil characteristics (e.g. texture) exists. The objective of this research was to assess the erodibility impact of soil disturbance. Undisturbed‐soil cores with dimensions of 45 cm (length) × 30 cm (width) × 10 cm (depth) were extracted while preserving their architecture. An A horizon corresponding to brown clayey subtropical oxisol soil from Southern Brazil was used for performing an experiment that involved simulation of 58–mm h−1 rain for 30 min. A total of seven replicate experiments were performed for each soil condition (i.e. undisturbed and disturbed soils). Results show that soil architecture deterioration had a larger impact on the involved soil loss than runoff. Further, soil structure failure did not affect the aggregate stability per se. Notably, the soil erodibility and loss were approximately 10 times larger under the disturbed‐soil condition than under the undisturbed‐soil condition (interrill erodibility: 4.30 × 107 and 4.39 × 106 kg s m−4, respectively; soil loss: 0.925 and 0.094 kg m−2, respectively). Overall, the intrinsic soil characteristics did not change; however the soil architecture deterioration considerably increased the erodibility.