In the elevated plus-maze test of anxiety the scores of control animals remain stable over repeated tests. However, a single prior exposure to the plus-maze renders an animal insensitive to the anxiolytic effects of chlordiazepoxide. This phenomenon of "one-trial tolerance" persisted even when the two trials were separated by as much as 2 weeks. It has previously been shown that the drug state of the animal on trial 1 is not important to the development of the phenomenon, but one-trial tolerance did not develop if a very high dose (75 mg/kg) of chlordiazepoxide was given on trial 1; it is suggested that this is due to the amnesic effects of the drug. The learning on trial 1 was not specific to a particular plus-maze and tolerance could be observed even when the maze on trial 1 was made from different material. The crucial experience on trial 1 was experience of an open arm of the maze. Whereas tolerance could be obtained as a result of a previous plus-maze experience, there was no evidence of an anxiogenic withdrawal response when rats were tested the following day undrugged. The phenomenon of one-trial tolerance is explained within our recently proposed two-factor theory of benzodiazepine dependence; it is suggested that one-trial tolerance provides a method for studying the mechanism underlying the development of tolerance to anxiolytic effects, independently from the mechanism underlying the development of withdrawal responses.