Abstract: Based on the influence of the aging and strain rate of a sloping soil mass on the strength and deformation of the soil mass, as well as the observations of many landslides cases, it has been demonstrated that the mechanism for formation of most landslides can be explained by the deep creep theory. Before a landslide occurs, the creep rate of a soil mass increases with an increase of the factors responsible for the landslide. According to this theory, ones can set up an empirical equation between deformation and time for predicting the time of failure of a landslide. Two earlier typical models are the Saito's model and Voight's model. The author has dug out some newly discovered facts hidden in the two models and developed a generalized prediction model, which can be used to forecast a landslide more accurately. Differing from the Saito's graphic solution method, the author has presented a more convenient technique for estimating the parameters in the equations. The presented method allows the data to be fitted with the presented model and expressed in terms of displacement, through non-linear estimation techniques using the SPRESHEET application software, in order to find values of the controlling parameters(A, B, m, t_f) suitable to represent the mechanical behavior of a large-scale landslide approaching the failure. A simple case study has confirmed the validity of both the presented generalized model and estimation technique through computing the deformation-time curve and defining the time limit or failure time value.