The ¹³⁷Cs  tracer technique has been widely used in the estimate of water-induced soil erosion rate representing a long term. There are two key problems calling for scientific solutions to practical utilization of this technique. One is the determination of reference value of ¹³⁷Cs content of soil having experienced neither erosion nor deposition, and the other is the establishment of the quantitative model between the amount of ¹³⁷Cs lost from the soil profile and the rate of soil loss. The primary existing quantitative model was reviewed in this paper, and their limitations were also revealed. These models fell into two categories, namely, empirical model and theoretical model, among which the mass-balance models are relatively reliable because they involved considerations of many factors related to the soil erosion process, such as annual fractions of ¹³⁷Cs fallout input, the grain size selectivity associated with soil erosion and sediment transport processes, surface enrichment,¹³⁷Cs  loss due to soil erosion prior to the tillage operation, and etc. The needs of some further improvements of existing models and some additional intensive works were also suggested in this paper.