Abstract: Aiming at the wind properties design of a bridge over a deep-cutting valley, 4 terrain models of different scopes are made for wind tunnel investigation, according to the scale ratio of 1:1000. The actual terrain areas covered by the four models are 25, 20, 9 and 1 square kilometers, resepectively. Moreover, the scope of the terrain models is normalized by the characterized dimension of the valley. In the wind tunnel experiment, the mean wind velocity, wind angles of attack and turbulence properties are measured with the wind direction that approximately parallel to the valley. The results indicate that, for the mean wind velocities and wind angles of attack, stable results are obtained when nondimensionalized distance from the bridge site to SE direction in the terrain model is greater than 2.2. For turbulence intensities at the bridge deck height, the two models whose nondimensionalized distance to SE direction is larger than 3.7 result in stable outcomes. However, the profiles of turbulence intensity differ significantly among the 4 models, which indicate that the scope of the models are not large enough as far as the profiles of turbulence intensity are concerned. Terrain model testsKG(-0.14mmhave been adopted extensively for the wind-resistant design of long-span bridges which ocated at the mountainous terrain. The investigation of this study indicates qualitatively that the scope of a terrain model for this kind of task should be cautiously selected.