This paper highlights central topics in language testing theory and practice that are relevant to the examination of modern language teaching and learning. Part 1 gives a time-lapse picture of the development of models of language competence in language testing, discusses the distinction between descriptive models and working models, and reviews the problem of distinguishing between underlying knowledge and performance. Part 2 deals with the operationalization of models in the form of tests, considering first the relationship between test method and the concept of communicative language testing and then addressing reduced redundancy testing and rating scales as measures of communicative competence. This second part also addresses the importance of reliability and validity, makes a case for standards of practice in language testing, and reviews language test equivalency across languages. Part 3 suggests four concrete implications for policy: all tests should be reexamined for validity in light of new language testing theory; testing should be matched up with communicative language teaching; language test batteries with crosslinguistic applicability should be constructed for all languages of the European Union; and research on the impact of tests on teaching should be carried out. (Contains 66 references.) (Author/NAV)