Many existing intonation studies tend to be unenlightening for three main reasons. First, they do not acknowledge that intonational variation may be functionally significant. By effectively ignoring functional variation among accents, investigators therefore run the risk of missing explanations for variation which may be theoretically important. The question of degree of perceptual relevance in intonational variation is not considered in detail here, although the basic assumption is that intonation is capable of performing a range of roles. Second, few studies attempt to provide a detailed explanation of the model they have used for analysing intonation. This means that there is no way of knowing whether, for instance, nuclearity in one accent is realised in identical ways in other accents, or even whether the concept of nuclearity is applicable in other varieties. Finally, existing accounts offer little information which is useful for developing intonational typologies. Clearly, this is a consequence of the failure to recognise variation and to provide an agreed analytic model. The present article addresses the shortcomings mentioned above and points towards a means of overcoming them by highlighting the need for a systematic phonological approach to intonation analysis.