Social-cognitive models propose that the other-race effect in face recognition is caused by different motivational tendencies when processing own- and other-race faces. More specifically, we tend to individuate own-race faces which facilitates face recognition, but racially categorize other-race faces which inhibits face recognition. This study tests whether a novel experimental manipulation aimed at promoting individuation or categorization encoding of faces moderates the other-race effect in an old/new recognition task. We found that categorization encoding eliminated the other-race effect when list length was short (Experiment 2) but not when list length was long (Experiment 1B). Inconsistent with social-cognitive predictions, individuation encoding failed to reduce the other-race effect, regardless of list length. We compare these findings with previous attempts to promote individuation and categorization encoding and suggest that the recognition benefits of individuation encoding might be more limited for faces that are most difficult to individuate i.e., other-race faces.