Music recommendation systems can observe user’s personal preferences and suggest new tracks from a large online catalog. In the case of context-aware recommenders, user’s current emotional state plays an important role. One simple way to visualize emotions and moods is graphical emoticons. In this study, we researched a high-level mapping between genres, as descriptions of music, and emoticons, as descriptions of emotions and moods. An online questionnaire with 87 participants was arranged. Based on the results, we present a list of genres that could be used as a starting point for making recommendations fitting the current mood of the user.