In this document, a review of traditional definitions and theories of metaphor suggests that they err in equating metaphors with comparisons rather than merely implying comparisons. Empirical research is then reviewed, revealing serious problems, particularly in developmental research. These problems include inadequate theories about the nature of metaphor, inadequate control over preexisting knowledge, and hasty conclusions that children cannot understand metaphors, drawn because in certain experimental conditions children do not select metaphorical interpretations. Related research on the comprehension of proverbs and analogies is discussed and recommendations for future research are made. These depend on a redefinition of metaphor and on adequate controls of preexisting knowledge, surface structure, and meaning. The approach suggested emphasizes the context-dependent nature of metaphors. Finally, it is argued that, even if metaphors can be transformed into comparisons, these comparisons are themselves nonliteral and, consequently, still need to be explained. (Author/AA)