In this experiment, strategies of pitch encoding in the processing of melodies were investigated. Twenty-six students who were highly trained in music and twenty-six who were less well trained were instructed to make recognition judgments concerning melodies after a 12-sec retention interval. During each retention interval, subjects were exposed to one of four conditions (pause, listening to an interfering melody, shadowing nonsense syllables, and shadowing note names). Both the standard and the comparison melodies were six-tone series that had either a high- tonality structure ("tonal melody") or a low-tonality structure ("atonal melody"). The results (obtained by Newman-Keuls method) showed that recognition performance for the musically highly trained group was severely disrupted by the note names for the tonal melodies, while it was disrupted by the interfering melody for the atonal melodies. On the other hand, for the musically less well trained group, whose recognition performance was significantly worse than that of the highly trained group even in the Pause condition, there were no significant differences in disruptive effects between the different types of interfering materials. These findings suggest that the highly trained group could use a verbal (note name) encoding strategy for the pitches in the tonal melodies, and also rehearsal strategies (such as humming and whistling) for the atonal melodies, but that subjects in the less well trained group were unable to use any effective strategies to encode the melodies.