Asthma is common, and some individuals are severely affected by it. Learned institutions have sought to provide a definition of 'severe asthma' to facilitate research and clinical care. This is a challenging undertaking given the difficulty in defining asthma and the lack of supportive evidence for a distinct severe asthma phenotype. In this review, we discuss the rationale for a definition of severe asthma and the relative merits of the sequential attempts that have been made to produce such a definition. The difficulty in disentangling control and severity is highlighted, as is the heterogeneity of phenotype in severe asthma, and potential for misclassification. We conclude that the search for a singular definition of severe asthma is problematic, though likely to continue. We suggest the alternative strategy of using classifiers with a specific aim related to symptoms, pathophysiology or service provision.