This study examined the salience of instrumentation over melody and harmonic accompaniment in identifying music excerpts. Musicians and nonmusicians were tested on match-to-sample recognition tasks. In the first task, subjects were asked to choose which resembled the model more: (a) identical melodies played with instrumentation different from the model, or (b) different melodies played with the same instrumentation as the model. In the second task, the same melodies were played, this time with harmonic accompaniments. Again, subjects were asked to choose between: (a) identical melodies played with different instrumentation, or (b) different melodies with the same instrumentation as the model. In the third task, subjects were asked to choose between: (a) a melody and its accompaniment played in a different instrumentation from that of the model, or (b) a melody and its accompaniment played in the same instrumentation as that in the model, but with the accompaniment played in the key that was dominant to the key of the melody. Ninety-five percent of the nonmusicians chose instrumentation over melody and harmonic accompaniment as the salient cue for recognition. Musicians always chose melody and harmonic accompaniment over instrumentation. These findings indicate that untrained listeners do not share or perhaps use the same cognitive schemata as trained listeners do. Further, the assumptions held by musicians that melody and harmonic accompaniment define the essential structure of much Western music, while instrumentation is cosmetic, may not be shared by nonmusicians.