The Data2Game project investigates how, and to what extent, the efficacy of computerised training games can be enhanced by tailoring the training scenarios to the individual player. The research is performed in close collaboration with serious-game developers at Thales/T-Xchange and with Brandweer Twente for the purpose of enhancing the training efficacy of firefighters. The research is aimed at achieving three concrete research innovations: (1) techniques for the automated generation of in-game narratives that are tailored to the learning needs of the individual player, (2) techniques for the automated modelling of players’ cognitive and affective states, based on in-game data and exhibited social signals, such that the training scenarios can be tailored to the individual player, and (3) validated studies on the relation of the player behaviour and game properties to learning performance. To achieve these innovations the project combines expertise in three disciplines: language technology (specifically: text generation), artificial intelligence (specifically: player modelling and personalisation techniques), and science of teaching (specifically: game based learning and skills assessment). The project is structured such that there is a close collaboration between the involved disciplines, and research outcomes will be implemented and validated directly in the actual target domain: the training of firefighters. The project will yield a new version of the Fire Brigade training game for Brandweer Twente at least once every year. Iterative user evaluation and empirical studies will be performed at the Virtual Reality Lab (University of Twente), DAF Technology Lab (Tilburg University), and at TRONED (Twente Safety Campus).