Abstract: The displacement-time curves published by Saito in 1960s imply landslide happening definitely in future, but actually, the gestation and development of landslide are of diversity and periodicity. For to walk out ties or limit from Saito's curves, the author puts forward that displacement-time curves of landsides can be classified as three types. They consist of change slowly to be stable, evolution in terraces and losing stability of slope. The stable type can transform to be terrace form in each other, and the latter transform to be failure type. In general, the type of losing stability can be not transform in reverse. Meanwhile, every type is interpreted with actual examples. Landslide occurs in the opening geological environmental systems. Landslide development is related intimately with its boundary conditions, constituents and structures, initial states and induced factors and their changes in time and space. Physical essence of landslide is out- of -balance of internal stress action in the slope which dynamic source is mainly from unbalance of gravity in slope. Losing stability of slope can be considered as it is into"automatic state" which means landslide would be happening inevitable. The type of evolution in terraces is susceptible with extraneous factor and the landslide is in the responding state or"passive state".The type to be stable landslide occurs only deformation adjustment in a local area which is in"weak dynamic state" unless it encounters extra-regular and extraneous factors. Because of the internal or external environments of landslides to be in the change endlessly, that scientists pursue"precise of early warning" of landslides would be impossible and unnecessary. The key problem is in the advance of service in disaster risk reduction.