This overview resumes first the historical development of avalanche forecasting in different countries of the world. It yields the most basic definitions like "avalanche danger" and presents the major problems of avalanche forecasting. Because avalanche forecasting is often compared with weather forecasting a comparison of these two forecasting problems is given. Due to the small-scale variations present in every snowcover, avalanche forecasting suffers from adequate data procurement mainly as far as the snow cover data are concerned. The necessary extrapolation from few point measurements imposes a serious restriction on the feasibility of avalanche forecasting. Also the longlasting problem of defining a reasonable "avalanche"-output-variable and its verification "a posteriori" is addressed. The various, over four decades concepted avalanche forecasting methods and models are described and valuated. Forecast-verification results are presented for the different approaches, whenever such results are available and declaretive. It clearly shows up that a detailed description or simulation of the snowcover on slopes seems mandatory if reliable forecasting results shall be obtained. Additionally an expert system approach, considering the heuristic aspects of forecasting, helps to overcome the complex problem of approximating the snow cover stability and fixing thereof definite danger degrees.