This thesis focuses on spot price forecasting and risk management in the Brazilian electricity industry. It is proposed a new methodology for the problem based on neuro- fuzzy systems and the dispatching and planning operation programs. The main advantage of the approach is to be able to get more informative spot price distributions than using the operation and planning programs alone. Furthermore, it allows Monte Carlo simulations or scenarios analysis as the forecasting system runs in less than 1 minute. The main variables which affect the spot price (inflow river, storage capacity of reservoir, among others) are included in the model. Even variables such as the interchange limits, without a well-defined time series and which could be important, could also be included because of the intrinsic characteristics of each fuzzy model. Comparisons with neural networks models are made. It is also presented the state-of-the-art in the market and politics modelling for the electricity market around the world, as well as some main concepts of the risk management.