In simple exchange experiments, participants have been reported to trade their endowments with similar objects less frequently than predicted by the neoclassical demand theory. Recent research has indicated that the experimental support for exchange asymmetry is fragile in alternative environments. In this article, we report an environment which is different from all other environments previously known in which exchange asymmetry is not observed. In this environment, participants knew at the time of making exchange decisions that they will have an opportunity to reverse their decisions. We find significant differences in exchange asymmetry in this environment, and an environment in which exchange decisions were final. Such differences suggest that anticipated regret theory provides an adequate explanation for exchange asymmetry which is often reported in the laboratory experiments. Our results underscore the importance of careful design for testing theories in the laboratory experiments.