Narratives often provide readers with opportunities to encode their preferences for particular outcomes. Our project examines some origins of such outcome preferences. For example, past literature suggests that readers tend to prefer positive outcomes for "good" characters and negative outcomes for "bad" characters. To extend this result we had participants read narratives featuring morally questionable protagonists and outcomes: in Experiment 1 the prologue of a novel ("Ghostman," Hobbs, 2013) and in Experiment 2 a short story ("Lamb to the Slaughter," Dahl, 1991/1953). We demonstrated that readers' individual assessments of characters as relatively likeable or moral affected the extent to which they rooted for outcomes favorable to those characters. Based on prior research, we also expected that readers' individual assessments would be related to their trait empathic concern. However, that prediction was not supported by the data. We also suggested that readers often have normative preferences for particular outcomes. In that context we explored the possibility that readers who were more highly transported by a narrative would root for those normative preferences. We obtained mixed evidence in support of that prediction. Finally, we measured readers' trait empathy and story enjoyment to replicate prior results on people's narrative experiences. Our results illustrate the complex origins of readers' outcome preferences.