Not all members of a category are equally good members; for example, a robin is generally considered to be a more typical member of the category ‘birds’ than an ostrich. Similarly, one vocal expression of, for example, happiness can be more typical than another expression of the same emotion, though both expressions clearly are perceived as ‘happy’. This chapter presents ongoing studies investigating the determinants of the typicality (graded structure) of vocal emotion expressions. In two experiments, separate groups of judges rated expressive speech stimuli (both acted and spontaneous expressions) with regard to typicality, ideal (suitability to express the respective emotion), and frequency of instantiation. A measure of similarity to central tendency was also obtained from listener judgments. Partial correlations and multiple regression analyses revealed that similarity to ideal, and not frequency of instantiation or similarity to central tendency, explained most variance in judged typicality. In other words, the typicality of vocal expressions was mainly determined by their similarity to ideal category members. Because ideals depend on the goals that people have, they can be independent of the particular category members that a person usually encounters. Thus it is argued that these results may indicate that prototypical vocal expressions are best characterized as goal-derived categories, rather than common taxonomic categories. This could explain how prototypical expressions can be acoustically distinct and highly recognizable, while at the same time occur relatively rarely in everyday speech.