This paper examines the problem of whether tonal representation should be considered segmental or suprasegmental. Woo's hypothesis that tone features are to be specified segmentally is discussed as well as Leben's statement that in some languages tone is suprasegmental. The following are criteria suggested as a basis for tonal representation: where tone is a feature of units larger than a segment, or if contour tones fall on single vowels, the representation should be suprasegmental; where phonotactic constraints refer to both tonal and nontonal constraints, the representation should be segmental. However, it is not yet clear what features are generally needed to represent tone either suprasegmentally or segmentally. Furthermore, there are indications that a viable theory of language will have to take into account varying types of tone languages which require further classification beyond that of requiring segmental or suprasegmental representation. (AM)