Abstract: Rainfall-runoff threshold is the minimum rainfall producing surface runoff and it is an important parameter for the research of runoff and sediment yield law. In this paper, the field artificial rainfall simulation experiments were carried out, and the rainfall threshold for different vegetation covers was obtained using the traditional regression method. Moreover,a nonlinear multi-parameter rainfall-runoff threshold model was established to analyze the effects of other factors on rainfall-runoff threshold. The results indicated that the rainfall-runoff thresholds of four different control measurements (bare slope,ryegrass slope, purple medic slope, and spring wheat slope) determined by the traditional linear regression method were 9.4, 23.6, 15.8 and 19.5 mm, respectively. Considering the effects of rainfall intensity, vegetation coverage and antecedent soil water content on the rainfall-runoff threshold, a nonlinear multi-parameter rainfall-runoff threshold model was established by using multiple regression analysis.By this model, the rainfall-runoff thresholds of the four different control measurements mentioned above (bare slope,ryegrass slope,purple medic slope, and spring wheat slope) were obtained as 13.4, 23.7, 18.8 and 19.7 mm, respectively. The nonlinear multi-parameter rainfall-runoff threshold model was tested by measured slope runoff data, and the results show that the calculated and measured thresholds match well, which proves that it is feasible to apply the nonlinear multi-parameter rainfall-runoff threshold model to ascertain the rainfall-runoff threshold.