A brilliant future is forecasted for Ambient Intelligence (AmI) systems. These comprise sensitive environments able to anticipate people's actions, and to react intelligently supporting them. AmI relies on decision-making processes, which are usually hidden to the users, giving rise to the so-called smart environments. Some of those envisioned environments include smart homes, health monitoring, education, workspaces, sports, assisted living, and so forth. Moreover, the complexity of these environments is continuously growing, thereby increasing the difficulty of making suitable decisions in support of human activity. Therefore, decision-making is one of the critical parts of these systems. Several techniques can be efficiently combined with AmI environments and may help to alleviate decisionmaking issues. These include classification techniques, as well as mathematical programming tools. In the first part of this work we introduce two AmI environments where decisionmaking plays a primary role: • An AmI system for athletes' training. This system is in charge of monitoring ambient variables, as well as athletes' biometry and making decisions during a training session to meet the training goals. Several techniques have been used to test different decision engines: interpolation by means of (m, s)-splines, k-Nearest-Neighbors and dynamic programming based on Markov Decision Processes.