Research has established that the encoding of events in memory can be impacted by our emotional state. However, when it comes to memory for music, remarkably little is known about the impact of emotional engagement with music on subsequently imagining that music. In collaboration with colleagues from the University of Leeds, a within-subjects musical imagery induction paradigm was used to investigate the relationship between emotional engagement when listening to music, and subsequent musical imagery. We hypothesised that Involuntary Musical Imagery (INMI) is more likely to occur, and to be more vivid, for music felt to be emotional than for affectively neutral music. Following pilot testing, we created stimuli by counterbalancing the pairing of emotionally neutral music tracks with positive, negative, and neutral film clips. Participants (N = 73) encountered these stimuli in an exposure phase, before completing a silent filler task. We then retrospectively asked about anyexperiences of imagining music during the filler task. Finally, a test of voluntary musical imagery accuracy (incorporating participants’ own music nominations) allowed us to test the hypothesis of greater imagery accuracy for music felt to be emotional than not. Binomial logistic analysis of INMI occurrence revealed that the most frequently imagined music came from the last stimulus presented, but also that music paired with the positive film significantly increased the odds of INMI occurrence. Neither INMI vividness, nor accuracy in the voluntary musical imagery task were affected by emotional valence. We provide new evidence of a link between positive emotion and subsequent INMI occurrence, with scope for further exploration of the role of emotional intensity as a factor contributing to musical imagery formation.