When benzodiazepines (BZs) supplanted barbiturates as a favored, safer treatment for anxiety and sleep disorders in the 1960s, the abuse liability and dependence potential of these drugs were little understood. Widespread recognition of the difficulty of stopping use of chronically taken BZs emerged through the popular press in the late 1970s, which resulted in reluctance to prescribe these otherwise clinically useful compounds. Evolution of the understanding of the biochemical basis for BZ effects in the 1980s and 1990s, coupled with regulatory emphasis on collection of data used in legal scheduling decisions, made possible a targeted search for drugs that would provide effective treatment for anxiety disorders in the absence of abuse liability or dependence potential. Compounds that have selective efficacy at subtypes of the gamma-aminobutyric acid type A receptor, are active in preclinical anxiolytic screens, but negative in preclinical studies of behavior relevant to evaluation of abuse liability appear to be one promising means for achieving this end.