We are interested in how digital games can be designed for learning in the affective domain. Our studies of how emotions are embedded in games and how games sustain affective learning involve observing gameplay and identifying recurring elements that we identify as design patterns. Design patterns help us think about the role of affect in play, what affect in games looks like, and the different ways affective learning might be achieved in educational and serious games. In this article, we describe and discuss several patterns related to understanding emotions, affective representation, and socioemotional interactions, which are essential components of affective learning. These patterns provide a language to conceptualize how affective learning might be designed into future game projects. To conclude, we discuss the development of a taxonomy of affective patterns to sustain socioemotional learning. We thus hope to stimulate the development of more human-oriented educational games in this domain.