Risk mitigation for debris flows at regional scale is a challenge. Early warning systems are helpful in depicting the time and the location of future debris flows so that emergency responders can act in advance before the disaster takes place. Herein we present a prototype real-time regional early warning system for rainfall induced shallow landslides and debris flows for the region of Catalonia (northeastern Spain). The model issues a warning level combining susceptibility information and real-time rainfall triggering conditions obtained from weather radar observations and forecasts. Susceptibility maps have been derived using a fuzzy logic approach and two input variables, terrain slope and land cover. These maps have been obtained using (i) grid cells of different resolutions, and (ii) physical catchments (of first order) as terrain units. Although high resolution grid-cell maps show a more accurate representation of susceptibility over the region, maps based on catchments are more intuitive and better characterize the area affected by future debris flows. Rainfall triggering conditions are assessed by means of probabilistic intensity-duration thresholds obtained from literature. Finally, we have validated the early warning system and tested its performance for some important events from the last ten years that were either monitored in specific catchments, or were reported in unmonitored catchments. In general, the system has been able to satisfactorily forecast the time of occurrence of most of the analyzed past debris flow events.