On the basis of evidence from general linguistic theory and psycholinguistic research, it is argued that the working hypotheses of the audiolingual habit theory are inconsistent with studies of human verbal learning tasks, and that transformation grammarians have shown convincingly that structuralist views of language are too narrow in scope. It is further argued that audiolingual precepts are intrinsically unsound and that modifying present approaches and techniques is not a viable solution. The cognitive code theory of language learning is seen as representing a truer picture of language and language acquisition. It is proposed that the cognitive code theory should be central to any method of foreign language instruction. Basic assertions underlying cognitive code theory are outlined, and the implications of their application to language teaching are examined. (MSE)