The research study aimed at the set-up of an Alert Prototype System (APS) functional to catchment characteristics and requirements, to support decision-making. Important side objectives are: a) the assessment of factors relevant to catchment response; b) assessment of hydrological parameters enhancing flood warning; c) determining procedures for flood warning and for the APS. A full APS has been successfully developed and tested in the Sesia river (Piemonte, north-west Italy) in collaboration with the Geological Survey of the "Regione Piemont" (in charge of flood management). This enables the achievement of the main objective of the research study and leads to interesting future developments. A first screen analysis of different catchments characteristics has been carried out to select the suitable catchments; the catchment choice depending on: a) amount and reliability of data; b) its significance among the different type of flood behaviour which can be observed in mountainous areas. Three catchments (Sesia, Tanaro and Toce) were selected at first, not only because floods frequently affect them but also for their dense population and the economic importance. This allows us not only hazard level assessment but also risk evaluation. The rainfall-runoff transformation module (NAM), the hydrodynamic module (HD) and flood forecasting module (FF) of the model MIKE11 have been used in the study. The choice has made from the authors' experience developed with this model, its reliability and the friendly use of the package. The numerical model has been extensively used in the Sesia and Tanaro catchments. In the Sesia the alert procedure has been developed with the use of NAM (for long and short period simulation), HD (for flow simulation in the river network) and FF (for real time flood simulation). In Tanaro river only the rainfall-runoff transformation and the hydrodynamic simulation have been carried out to represent synthetic events on the catchment, to define different contribution from tributaries in relationship with different rainfall patterns and to assess general flood behaviour. The present study shows how it is possible to develop a real time forecasting system for a natural mountain catchment taking into account of both modelling techniques for hydrological and hydrodynamic simulation and threshold values for defining vulnerability and risk parameters. The study underlines also the difficulties normally encountered when studying and forecasting floods in mountainous areas. This is particularly evident in Piemonte where climatic, physiographic and morphological characteristics vary very much within single catchment.