Masked priming research with late (non-native) bilinguals has reported facilitation effects following morphologically derived prime words (scanner – scan). However, unlike for native speakers, there are suggestions that purely orthographic prime-target overlap (scandal – scan) also produces priming in non-native visual word recognition. Our study directly compares orthographically related and derived prime-target pairs. While native readers showed morphological but not formal overlap priming, the two prime types yielded the same magnitudes of facilitation for non-natives. We argue that early word recognition processes in a non-native language are more influenced by surface-form properties than in one's native language.