Auditory and audiovisual training have been shown to be successful in increasing the discriminability of non-native phonetic contrasts in second-language learners. The purpose of this study, which trained the English /l/-/r/ contrast with Japanese learners of English, was to investigate training effectiveness using visual stimuli alone. The study evaluated whether training with visual cues leads to (a) increased discriminability of the phonetic contrast, (b) an increase in visual influence in phonetic labelling, and (c) cross-modal effects in audiovisual or auditory speech perception. Visual-alone training was successful in increasing the discriminability of the /l/-/r/ contrast in visual and audiovisual test conditions but there was no carry-over to the auditory condition. There was also evidence of an increase in audiovisual advantage (AV>A) in the training group and of good generalisation to unknown words by the same speakers and to nonsense words by unknown speakers.