Abstract: Hillslope is the fundamental unit of hydrological response in mountainous small watersheds, and elucidating the nonlinear threshold characteristics of its runoff generation process is essential for understanding storm flood formation mechanisms and improving flash flood early warning capabilities. This study took a typical flash flood-prone small watershed in the Qinba Mountains as the study area, and employed a combination of indoor simulated rainfall experiments and field runoff plot observations to systematically investigate the synergistic control mechanisms of multiple factors on surface runoff initiation time, including rainfall intensity (RI), slope gradient (S), soil thickness distribution pattern (STDP), and initial soil moisture content (θ_ini). A theoretical model for predicting surface runoff initiation time was developed, and an integrated threshold index was proposed from the perspective of water balance.The results demonstrated that: (1) Rainfall intensity, slope gradient, soil thickness, and STDP exerted nonlinear synergistic effects on surface runoff initiation time. Rainfall intensity was the dominant controlling factor, with runoff initiation time decreasing sharply as a power function with increasing intensity. The slope effect exhibited strong rainfall intensity dependence, being pronounced under low-intensity conditions but diminishing under high-intensity conditions. Soil thickness determined the baseline response level by regulating water storage capacity; runoff initiation time in thick soil layers was 35%～50% longer than in thin layers. STDP primarily modulated the effects of rainfall intensity and slope by altering subsurface flow pathways. Initial soil moisture content was significantly negatively correlated with runoff initiation time. Variance analysis revealed that the interaction between rainfall intensity and slope contributed 19%～20% of the total variance, while the interactions of STDP with rainfall intensity and slope jointly accounted for 30%～35%. (2) The theoretical model developed based on multi-factor synergistic mechanisms performed well, achieving R² values of 0.91～0.92 and Nash-Sutcliffe efficiency (NSE) values of 0.89～0.92 in laboratory validation, and demonstrated reliable performance in field runoff plot validation within the Guanshan River Basin. (3) Based on the classical three-stage runoff generation theory, the integrated index “cumulative rainfall + initial soil moisture content” (CR+ASM) clearly characterized the critical thresholds of the three stages, water storage, slow runoff, and fast runoff, and effectively eliminated the influence of soil thickness differences, reducing threshold disparities between thick and thin soil layers to within 2.32 mm. The theoretical model based on this integrated index achieved excellent simulation accuracy (R²=0.99, NSE=0.99) and effectively suppressed simulation biases under extremely dry antecedent conditions, reducing relative errors from −20.0% to −26.2% down to −6.39% to −9.13%. This study revealed the multi-factor synergistic nonlinear threshold mechanisms of hillslope surface runoff generation. The developed theoretical model for runoff initiation time and the CR+ASM integrated discriminant index provided important theoretical foundations for accurate storm flood simulation and flash flood disaster early warning in mountainous small watersheds.