Three experiments that form an empirical basis for discussing the cognitive reality of hierarchic structure in music are reported. The first experiment showed evidence of listeners' ability to match a performed reduction of an extract of tonal music to the piece of music from which it was derived. A second experiment showed that this choice of reduction could not be attributed to the relative " coherence" of reductions. These two experiments provide evidence for the internal representation of tonal music in terms of a hierarchy of events such as that proposed by Lerdahl and Jackendoff ( 1983). In a third experiment using atonal music, subjects were less successful in choosing as the best reduction that which resembled the extract at higher levels of the structural hierarchy. Thus there is no evidence for the perception of a hierarchy of events in atonal music of the sort proposed by Lerdahl (1989). This empirical work therefore suggests that whereas the tonal system allows events within a tonal work to be heard within a strict hierarchy, no such hierarchy exists for atonal music. This finding has two main implications. First, a new conception of the term "prolongation" is needed if it is to apply to atonal music. The lack of a pitch hierarchy means that atonal events are unable to "stand for" other events in the way that tonal events are, and it is this action of standing for that allows prolongation to occur. Second, if as this research suggests, atonal music is not perceived in terms of a hierarchic structure, then another approach may be to investigate associational properties of the music and the role that these play in the formation of a structural representation.