Abstract. Current findings suggest that human-computer interaction following the basics of human-human interaction, in which emotions play a critical role. We performed a set of consecutive experiments in a laboratory environment for identification, recognition, visualization and interactive computing of affective states within a gaming application framework. To start our research and to get a data base we analyzed more than 90 test hours of user tests using rating-scales and physiological measurements. As results we provide (1) a mini-game with extra features for the induction and obtaining of affective states, (2) integrated data mining methods with recognition rates up to 70 percent, (3) different kind of visual representation of recognized emotions, and (4) an architecture for control of an affective game.