Hydrological models are a fast and low-cost option to estimate surface runoff in a rainfall event in ungauged basins. Prior to their use, the models must be evaluated; their results, if reliable, are useful in def ining sound policies for rational management of the water resource. The purpose of this research was to evaluate the degree of f it of the numerical curve runoff Model of the extinct U.S. Soil Conservation Service (SCS-CN) and the Green-Ampt (GA) inf iltration model to reproduce direct runoff hydrograms in ungauged basins. These models were integrated into a hydrological model structured in HEC-HMS software. Seven well-def ined rainfall events and one prolonged event with intermittent precipitation occurring in the Chapingo River Basin, Mexico, were chosen. In the analysis of the set of eight events, the SCS-CN model estimated the total runoff volume better (RMSE of 5430 m3), while the GA model estimated the peak flow better (RMSE of 0.72 m3 s-1). The GA model best estimated the hydrograms of three well-def ined rainfall events whose runoff volumes were greater than 8.31 × 103 m3 and Qp ≥ 1.93 m3 s-1 with RMSE of 0.14 to 0.43 m3 s‑1, MRE of 0.20 to 0.28 and Nash-Sutcliffe eff iciency of 0.92 to 0.95. For the other four well-def ined events with runoff volumes lower than 6.35 × 103 m3 and Qp ≤ 1.44 m3 s-1, both models estimated the hydrograms identically and without effectiveness with a Nash-Sutcliffe index of -1.38 to -7.34. Estimation of the hydrogram of the prolonged event was identical in both models with an RMSE of 0.07 m3 s-1, an MRE of 0.35 and a Nash-Sutcliffe eff iciency of 0.71. The hydrogram of the prolonged event was identical in both models, with an RMSE of 0.07 m3 s-1, an MRE of 0.35 and a Nash-Sutcliffe eff iciency of 0.71.