The current study investigates how infants categorize human compared to ape faces. Nine‐month‐old infants were presented with priming stimuli related to human (N = 24) or ape (N = 25) face targets on different levels of categorization. Event‐related potentials were recorded during a passive‐looking rapid repetition paradigm. In a within‐subjects design, priming effects of the same faces, different faces from the same basic‐level category, different faces from the other basic‐level category (human/ape faces), and house fronts were examined. Human and ape faces were first categorized on a superordinate level (“faces”), as indicated by enhanced P1 amplitude and reduced P1 latency for faces primed by any faces. Then, human and ape faces were categorized on a basic level. N290 amplitude and latency were larger for human and monkey targets primed by human faces. Neither human nor ape faces were categorized on the individual level.