This thesis is divided in 3 chapters: first chapter refers to the evaluation of forecasts accuracy and ways to improve it, the second one presents the problem of accuracy evaluation of a macroeconomic aggregate based on two different strategies of forecasting. Finally, the last chapter develops the problem of building forecasts interval taking into account the state of the economic. Taking into account that inflation rate follows an AR(1) process for USA, we found out a new method to generate forecast intervals that take into account the state of the economy. The problem of forecast accuracy is related to the uncertainty that characterizes each decision process. I consider that the variables' aggregation is an important source of uncertainty in forecasting that is not specify in literature till now. Therefore, I propose the introduction of variables' aggregation among the sources of forecasts uncertainty.