In this chapter, estimation of distribution algorithms will be described. These algorithms belong to the evolutionary computation field and are characterized by evolving a population of candidate individuals as solutions of the optimization problem estimating at each generation the joint probability distribution of the selected individuals and then sampling new individuals from that distribution. Thus how to model (and sample from) that joint distribution is an important issue that in this chapter will be analyzed. The specificities of these algorithms when dealing with multimodal, multiobjective, or dynamic optimization problems will also be overviewed. Then parallelization and hybridizations with other optimization heuristics will be presented. Next, applications of the algorithms will come in the last two sections. On the one hand, the use of these algorithms to solve real-world optimization problems in biomedicine, bioinformatics, energy, vehicle routing, and scheduling will be shown. On the other hand, applications in different machine learning tasks, such as supervised classification, clustering, and Bayesian networks, will be discussed. Finally, the conclusions will round the chapter off.