We analyze the production of [±voiced] stops in French as a foreign language by multilingual learners who speak Russian or Turkish as a heritage language along with their dominant language, German. Control data produced by monolingually raised German learners are taken into account. It is shown that the bilingual learners perform more target-like than the monolinguals for the voiceless stops /ptk/, but not for their voiced counterparts /bdɡ/. This suggests that the (non-)aspiration of voiceless stops is perceptually more salient for the learners than the presence or absence of pre-voicing in the realization of voiceless stops. We interpret our overall results as an example of an at least partial multilingual advantage in foreign language learning.