Traditionally, statutory provisions prohibiting direct discrimination have employed a test of less favourable treatment. This test is controversial since it requires the use of a comparator, which is widely regarded as giving rise to a range of serious problems. It is commonly assumed, both by academics and legislators, that reliance on a comparator, and the problems to which such reliance gives rise, can be avoided by employing a test of unfavourable treatment instead of a test of less favourable treatment. In this article, we subject this assumption to critical scrutiny. We acknowledge that, on what is probably the most common understanding of the test of unfavourable treatment, employing that test does avoid the need to rely on a comparator. However, we argue that this understanding renders the test of unfavourable treatment radically over-inclusive. We then consider alternative approaches to understanding the test of unfavourable treatment, and investigate whether these approaches avoid the need to rely on a comparator whilst also avoiding the over-inclusiveness problem. We argue that this depends, ultimately, on what the value is that underlies prohibitions on discrimination.