The present research tests Japanese listeners' perception of Italian liquids under the framework of the Perceptual Assimilation Model (PAM), a model of non-native speech perception. The study examines the discrimination and the identification of Italian liquids (/l/, /r/, /ll/, /rr/ and /ʎ/), while comparing the results with those of replicated perception experiments of Spanish liquids (/l/, /ɾ/ and /r/). Japanese listeners showed poor discrimination performance on Italian /l/-/r/ and very good discrimination performance on the other Italian liquid contrasts (/l/-/ll/, /l/-/rr/, /ll/-/r/, /ll/-/rr/, /r/-/rr/, /l/-/ʎ/, /ll/-/ʎ/, /r/-/ʎ/ and /rr/-/ʎ/). The listeners perceptually assimilated Italian /l/ and /r/ to Japanese /r/, Italian /ll/ to Japanese /Qr/, Italian /rr/ to Japanese /rur/, and Italian /ʎ/ to Japanese /rj/. The findings indicate that PAM accounts for the perception of Italian liquids by Japanese listeners.