In the following study, we reexamine the significance of the fact that subjects can solve randomly generated analogies. This phenomenon has long been demonstrable and has usually been viewed as evidence for the tremendous flexibility inherent in language. Such demonstrations have been used to question the idea that any type of psychologically interesting structure can be recovered from the study of language. In this article, the results of four random-analogy studies were examined in an effort to understand how subjects can find meaning in random analogies. The results indicate that subjects make use of a small number of relational concepts in imposing a structure on this seemingly unstructured task. This study, and the work of others, suggests that these relational concepts play a central role in metaphor, analogy, and cognition in general. (Michael G. Johnson and Tracy Henley)