Taxonomies have been developed as a mechanism for cyber attack categorisation. However, when one considers the recent and rapid evolution of attacker techniques and targets, the applicability and effectiveness of these taxonomies should be questioned. This paper applies two approaches to the evaluation of seven taxonomies. The first employs a criteria set, derived through analysis of existing works in which critical components to the creation of taxonomies are defined. The second applies historical attack data to each taxonomy under review, more specifically, attacks in which industrial control systems have been targeted. This combined approach allows for a more in-depth understanding of existing taxonomies to be developed, from both a theoretical and practical perspective.