Abstract: In the double yield surface model proposed by Lade, the projection of shear yield surface and that of volume collapse yield surface are significantly different on plane. This deficiency is discussed theoretically and experimentally, then a modified Lade volume collapse yield surface is suggested, which is further examined by testing results. Meanwhile, another reasonable volume collapse yield surface is also recommended. Thus stress and strain characteristics of soil mass could be predicted reasonably by the combined Lade shear yield surface with the modified Lade collapse yield surface or with the recommended collapse yield surface.