The thesis of grammatical deviance, widely accepted for many years, holds that a string is interpreted as a metaphor only if, when interpreted literally, it would be in some way grammatically deviant, absurd, etc. The strongest version of this thesis relies strictly on grammatical deviance as a criterion for metaphorical interpretation. Some assumptions shared by both critics and supporters of this thesis are examined and questioned. Both approaches have failed to recognize the proper role of grammar and extralinguistic context in metaphor identification and interpretation. An attempt is made to determine the true contribution of grammar to the interpretation of metaphor. Some important aspects of M. Beardsley's theory of metaphor are examined in detail. (CSA)