This paper aims to study whether training with gestures benefits L2 phoneme acquisition fromboth a production and perception perspective. In the production study, Dutch learners ofSpanish received pronunciation training for the phonemes /u/ and // in one of four conditions:audio-only, audio-visual, audio-visual with pointing gestures, or audio-visual with iconicgestures. Results show that in general, gestural training benefits L2 phoneme acquisition, butdifferent gestures benefit the acquisition of different phonemes, possibly depending on theircomplexity. The perception study, in which L1 speakers of Spanish judged the L2 Spanishmaterial on accentedness and comprehensibility, corroborate the findings from the productionstudy: Including visual information in training generally lowered the perceived accentednessand increased the perceived comprehensibility of speech, but the type of phoneme matters.Together, these studies suggest that gestural training can benefit L2 phoneme acquisition, yetcertain gestures work better for certain phonemes than others.