Vast efforts have been made to develop novel anxiolytic drugs that improve on those that target the GABA (γ-aminobutyric acid)–benzodiazepine system, but promising results in rodents have rarely translated into effectiveness in humans. Griebel and Holmes analyse the major trends from a database of published preclinical studies on novel anxiolytic agents in the past 50 years, highlight issues that may have hampered progress and offer recommendations to improve anxiolytic drug discovery.