In this chapter, we present a robot soccer system that learns by imitation and by experience. At first the robots do not know anything but after observing saved games of other teams, they start improving their knowledge learning new states and actions. This is called learning by imitation. They validate learned actions by testing them several times, which called learning by experience. Repeating this process allows that robots can continuously improve their performance. This strategy emulates the way humans learn. © Springer-Verlag Berlin Heidelberg 2007.