Based on research carried out by the international scientific community on climate variability and long range forecasting along with the developpment of comprehensive ocean data sets with easy and regular access made available by various international and national centers such NCEP in US and the UK meteorological office as well as the existence of a regular monitoring of the ocean variability such as one made at the European Center for Medium Range Forecast - ECMWF- and the Bureau of Meteorology in Australia, it was recommended from various past workshops and conferences related principally to WMO world climate research program and the CLIPS projects the establishement of a capacity building approach for national meteorological services in tropical to seek the potential use of climate predictions in various socio-economic sectors. Following these recommendations the author explored the existing theories and knowledge in climate variability and prediction during his duty as a head of climate unit at the African Centre of Meteorological Application for Development to seek for an operational system for climate prediction based on empirical model. The approach is based on statistical basic knowledge and empirical models using Sea Surface Temperature (SST) indices as predictors and seasonal rainfall indices as predictands. The approach provides steps for definining climatic zoning, teleconnection analysis based on pearson correlation analysis with various SST indices, empirical modeling and validation on past records. The approach was applied during several training workshop organised within Africa. It was particularly found that during recent ENSO events, 1998,1999 and therafter a real time predictability assessment was encouraging and has been therefore particularly useful in issuing seasonal forecasts. Inspite of the encouraging results of this approach in Africa, it has been applied for other regions of the world such as during a six weeks workshop organized by the cooperative institute for mesoscale studies at the Oklahoma univesity in United States in october 1999, in which several participants from Latine America produced predictability with good skill based on historical data validation. The paper presents as well a PC based interactive software which has been developped by the author to provide an early warning system for drought mitigation and flood anticipation for decision makers in west Africa and northern Morocco.