Music can evoke powerful emotions in listeners. An open question is whether instrumental music further conveys extra-musical meaning. We conducted an online study in which a large cohort of participants (N = 121) listened to twenty 15-second-long excerpts of unfamiliar instrumental soundtrack music and identified imagined movie scene elements (e.g., scene brightness, character role) and emotions (e.g., happiness, sadness) evoked by each excerpt. We systematically investigated how acoustic features of instrumental soundtrack excerpts (e.g., tempo, pitch) and the emotions that they evoked contributed to mental imagery. Our results reveal similar contributions of all acoustic features to the imagination of elements of movie setting (e.g., lighting, temperature), but distinct and mostly non-overlapping contributions to the imagination of elements that describe actions, objects, and characters. Moreover, we found that emotions with negative valence fully mediate the relationship between acoustic features and movie scene elements. Our findings carry implications for cognitive neuroscience and music cognition research as well as for the study of media entertainment.