Context aware recommender systems (CARS) adapt to the specific situation in which the recommended item will be consumed. So, for instance, music recommendations while the user is traveling by car should take into account the current traffic condition or the driver’s mood. This requires the acquisition of ratings for items in several alternative contextual situations, to extract from this data the true dependency of the ratings on the contextual situation. In this paper, in order to simplify the in-context rating acquisition process, we consider the individual perceptions of the users about the influence of context on their decisions. We have elaborated a system design methodology where we assume that users can be requested to judge: a) if a contextual factor (e.g., the traffic state) is relevant for their decision making task, and b) how they would rate an item assuming that a certain contextual condition (e.g., traffic is chaotic) holds. Using these evaluations we show that it is possible to build an effective context-aware mobile recommender system.