This study assumed that the relation between readers’ and characters’ group identity would influence narrative empathy and thereby the impact of a short story. Three social psychological models (infrahumanization, mentalization and linguistic inter-group bias) were used to test these assumptions. Methodologically, a narrative recall paradigm was used, based on the assumption that narrative recall carries also the experiential aspects of the text processing and thereby enables a fine grained analysis of meaning construction. The study also measured liking of the story as an impact variable and the strength of national identification as a moderator variable. Results only partly supported the assumptions. Whereas empathy and liking are strongly correlated, Hungarian subjects overall did not feel more empathy with the characters of the Hungarian story version, did not like more this story, did not assign more secondary emotions to the Hungarian characters, and did not describe positive behavior of the Hungarian characters more abstractly then happened in the case of the “Slovak story” version.