In recent years, physically-based numerical models have frequently been used in the framework of early-warning systems devoted to rainfall-induced landslide hazard monitoring and mitigation. For this reason, in this work we describe the potential of SLIP (Shallow Landslides Instability Prediction), a simplified physically-based model for the analysis of shallow landslide occurrence. In order to test the reliability of this model, a back analysis of recent landslide events occurred in the study area (located SW of Messina, northeastern Sicily, Italy) on October 1st, 2009 was performed. The simulation results have been compared with those obtained for the same event by using TRIGRS, another well-established model for shallow landslide prediction. Afterwards, a simulation over a 2-year span period has been performed for the same area, with the aim of evaluating the performance of SLIP as early warning tool. The results confirm the good predictive capability of the model, both in terms of spatial and temporal prediction of the instability phenomena. For this reason, we recommend an operating procedure for the real-time definition of shallow landslide triggering scenarios at the catchment scale, which is based on the use of SLIP calibrated through a specific multi-methodological approach.