Readers rely on descriptions of characters to generate expectations for future story events. However, readers also have preferences for how they want those events to unfold. Often, what texts imply about how characters will behave and what readers want characters to do converge on similar story events and narrative descriptions. But what are the processing consequences when expectations and preferences suggest competing possibilities? In three experiments, we explored this question utilizing short narrative texts. Each text included information designed to establish positive, negative, or neutral preferences toward characters, as well as behavioral descriptions that supported particular positive or negative character traits. In Experiments 1 and 2, when the valences of reader preferences and implied traits matched, participants overwhelmingly judged characters as likely to possess those traits. With mismatches, though, those judged likelihoods decreased in systematic ways. In Experiment 3, we observed that matches between preferences and implied traits also influenced reading times for story outcomes. These results outline how the inferences that guide narrative comprehension are influenced both by the descriptions that authors provide about characters and events, as well as by the emerging desires that readers develop for those characters and events.