Since the 1997 election of the Labour Party to political power in the UK the foci for educational change have been widespread. One area that has received particularly intense scrutiny is that of teacher activity. In particular, the profession has seen a marked rise in the identification of "best practice". As a term "best practice" has entered the parlance of English educational policy to describe that which seemingly has "official" approval. This paper uses a social constructionist perspective to consider how increases in pupil attainment on national tests are currently used to demonstrate better pupil learning. Specifically, it identifies that the use of such data to describe the plausibility, veracity and legitimacy of teaching before the test as "best practice" is questionable. In so doing, the critique argues that "best practice" confers and retains legitimacy due to its self-perpetuation within the discourse of performance. The paper concludes by offering three areas for further research and debate. (Contains 1 note.)