Anxiety is a complex state that includes a broad range of classified symptoms. Anxiolytic treatment has been dominated by the use of benzodiazepine drugs (BZs), which is often prolonged. However, the multiplicity of anxiety states and the lack of BZ specificity preclude the potent targeting of the drugs to different states, and point to the problem of drug dependence. Furthermore, animal models of anxiety give different results when challenged with apparently similar drugs. The search for alternative drugs with a higher specificity has led to 5-HT receptor subtypes. In this review, Simon Green describes how recent research might have dual benefits: in improving drug therapy of anxiety in humans, and in analysing and categorizing more clearly the concept of anxiety itself.