Large-scale assessments have become a basic national policy for educational improvement encouraging standards, decentralisation and school accountability. The current study focuses on the pedagogical dimension of large-scale assessments, examining its uses as a policy instrument for effecting pedagogical change. The paper presents and discusses the case of the Israeli NLSA (national large-scale assessment) regime--the Meitzav (Hebrew acronym for: Growth and Effectiveness Measures for Schools) tests. Although it aimed to design a low-stakes testing regime, its implementation was a top-down procedure which, in practice, restricts principals' and teachers' autonomy. Using qualitative and quantitative methods, the findings showed that the Meitzav test results are barely used as a means leading to pedagogical change to improve learning. Teachers considered these tests an unreliable assessment tool that in the main, does not reflect the school curriculum or student learning, while producing a high level of pressure on the teaching routine. In consequence of the Meitzav test results, the most common pedagogical change in practice chosen by teachers was: "teaching to the test". Other pedagogical changes following the Meitzav were implemented to a minor extent. Policy implications are discussed.