Scotland has a system of national surveys of achievement intended to provide information for educational policy and practice. Testing of middle school students in the areas of mathematics, reading, English, and science, is designed to provide national information as opposed to information about individual pupils, schools or local authorities. This paper examines an early series of national surveys (1932, 1953 and 1963) and finds that (1) if surveys are to produce an impact on educational policy, short-term focused studies rather than a long-term series of national surveys addressed to general issues are indicated; (2) knowledge of specifics relevant to particular local circumstances are usually sought; (3) national tests will inevitably have curriculum backwash and will involve pressure towards a centrally determined curriculum; and (4) research that is to have an impact on classroom practice must be teacher tested and based upon careful research and development in classrooms. It is concluded that surveys are unlikely to address the crucial issues of the time and are likely to be misleading to any individual practitioner. (PN)