The aim of the proposed grant project is to amplify results obtained in the previous projects (especially project 402/05/0115 evaluated as excellent, and the current project 402/08/0107). Attention will be primarily focused on various approaches for description of the system dynamics, long decision horizons in multistage stochastic programming problems and their connections with dynamic programming methods, as well as on adequante evaluation of the models. For handling concrete problems it is important to obtain credible data and to employ full information including development of the considered system over time. This ask for formulation of more complex models and to further development of approaches to data preprocessing, input generation and output analysis bearing in mind the required robustness of decisions, and to verification of the new model using multimodelling. Doctoral students will be also involved in the project. The project will be coordinated with universities and academic institutions from abroad (University of Bergamo, Italy; University of Amsterdam, Department of Economic and Business, Amsterdam; Humbolt University, Berlin BRD; Universidad Autonóma Metropolitana - Iztapalapa, Mexico City).