Comprehension is integral to enjoyment of media narratives, yet our understanding of how viewers create the situation models that underlie comprehension is limited. This study utilizes two models of comprehension that had previously been tested with factual texts/videos to predict viewers' recall of entertainment media. Across five television/film clips, the landscape model explained at least 29% of the variance in recall. A dual coding version that assumed separate verbal and visual representations of the story significantly improved the model fit in four of the clips, accounting for an additional 15–29% of the variance. The dimensions of the event‐indexing model (time, space, protagonist, causality, and intentionality) significantly moderated the relationship between the dual coding model and participant recall in all clips.