Although the fallout radionuclide ¹³⁷Cs technique has been widely used to provide soil redistribution estimates in recent decades, some researchers have recently challenged the usefulness of this technique. Thus, it is necessary to evaluate the usefulness of the 137Cs technique to study soil redistribution, especially in areas that suffer from severe soil erosion and where the ¹³⁷Cs inventory distributions vary. This study
analyses the abundance of ¹³⁷Cs deposited in sediment trapped by a check dam over 31 years in a small agriculture catchment in the Loess Plateau, which has experienced some of the most extensive agricultural soil erosion in the world. The ¹³⁷Cs inventory and soil loss were evaluated, and the results indicated that the sediment deposition rate behind the check dam was approximately 38 cm/year. The specific soil
erosion rate of the catchment is estimated to be 178 t ha^-1 year^-1. A geochemical properties fingerprinting method indicated that the ratio of
the inter-gully erosion from cropland to gully erosion was 0·46:0·54 during the investigated period of 31-year period. Our results suggest that
the ¹³⁷Cs technique cannot be directly used in the study catchment. Gully erosion is the main type of soil erosion that occurs in the loess catchment, and the amount of gully erosion changes from year to year. Therefore, the abundance of ¹³⁷Cs cannot be used to estimate soil redistribution where gully erosion is severe. However, after calibration, the inter-gully soil loss was consistently related to ¹³⁷Cs abundance,
which means that the ¹³⁷Cs technique can be used at the plot or slope scale on the Loess Plateau. Our results also reveal the particle-size selectivity of soil detachment and sediment deposition during runoff erosion events. The average ¹³⁷Cs content of the fine group couplets
is 63% greater than the ¹³⁷Cs content of the coarse group. When the studied section spans more than 15 years, the ¹³⁷Cs inventory is consistent (R² = 0·95, p < 0·01) with soil loss, which means that the selectivity may be offset over a long period. Our results confirm that the ¹³⁷Cs method requires careful planning when applied at the catchment scale.