Various studies have built models, using aggregate box-office data, to predict the contribution of a motion picture’s features to its theatrical demand. But such an approach fails to represent the heterogeneous influence of movie features on demographic groups and is unable to assist market-segmentation decisions. We propose and illustrate a new approach for modeling the appeal of movie features to market segments via the use of appropriate individual-specific data and canonical correlation analysis. Specifically, through demographically detailed movie-attendance data available in Spain, we build a model of how movie features influence the demographic composition of audiences. Via a canonical correlation analysis, we identify four dimensions underlying the relationships between several movie features (country of origin, genre, objectionable content, stars, promotional effort, and critical evaluations) and audience demographics (gender, age range, presence of children, education, social class, and size of municipality). These dimensions represent the strong pairings between four moviegoer demographic profiles and four movie-feature profiles. Our approach can potentially aid in segmentation and green-lighting decisions by matching movie features with the most relevant segment-specific preferences.