The present study examines the effects of training native Japanese speakers in the production of American /r/ using spectrographic visual feedback. Within a modified single-subject design, two native Japanese participants produced single words containing /r/ in a variety of positions while viewing live spectrographic feedback with the aim of producing /r/ with a third formant (F3) frequency of less than 2300 Hz (upper threshold for identifiable /r/). Feedback was gradually reduced to promote independent production and monitoring. Both participants showed improvement in /r/ production in trained and untrained single words as indicated by percentage increases and effect sizes. Blind ratings by independent experts indicated significant /r/ production improvements during the reading of the Rainbow Passage. Perceptual distinction of /r/ and /l/ in minimal pairs also showed an upward trend. These findings suggest that spectrographic visual feedback is a promising method for training /r/ to Japanese-speaking English language learners.