In the late 1950s, researchers at Hoffmann-La Roche discovered that benzodiazepines were able to cause changes in animal behavior that resembled those caused by barbiturates and meprobamate. Shortly thereafter, chlordiazepoxide was introduced ino clinical practice as an anxiolytic, and other benzodiazepines soon followed. They were greatly favored over meprobamate and barbiturates not only because of their wide safety margin—death from benzodiazepine overdose alone is extremely rare—but also because benzodiazepines were not thought to produce dependence except following prolonged administration of very high doses. During the 1980s, this latter belief underwent considerable revision (see File, 1990 ), and there is now considerable interest in the development of drugs with less dependence-producing potential for the treatrnent of anxiety disorders.