A model of MultiSource Information Fusion (MSIF) is proposed. It expands the application of proven MSDF techniques to diverse problem areas. This model allows for a unified framework clearly distinguishing processing functions from methods dealing with partial, uncertain, and imprecise information. The concept of computational intelligence provides for a holistic approach to design and integration of methods and algorithms for information fusion. We describe the application of computational intelligence to the fusion of data and information in two studies of early warning. The emphasis is on the power of soft-computing methods in designing early warning architectures pertinent to forecasting events in complex dynamical systems.