This paper reports on a mid-point evaluation of the Gauteng Primary Language and Mathematics Strategy (GPLMS), an innovative large-scale reform designed to improve learning outcomes. Using data from universal testing of all learners in 2008 on a provincial systemic evaluation, and data from the 2011 and 2012 Annual National Assessment (ANA) test, this paper addresses the key research question, namely whether the GPLMS is effective in closing the gap between performing and underperforming schools. Given the evidence we have presented of an instrument effect, namely that various versions of the ANA may not be strictly comparable, no definitive conclusions can be drawn about the effectiveness of the GPLMS.