This participative case study took as its starting point a current “problem”, namely teachers’ concerns about assessment in self-managing primary schools. Prompted by teachers’ stress and confusion about their assessment role, I investigated how, in the 1990s, New Zealand teachers responded to multiple competing discourses of assessment, and the effects these discourses had on their teaching practices. Central to this thesis is the notion that, since compulsory schooling began in 1877, teachers’ assessment practices have been both constituted by and constitutive of dominant educational and assessment discourses. Taking account of the Foucauldian stress on discontinuity, this thesis argues that traces of teachers’ assessment practices can be found in earlier dominant discourses. Evidence is presented to show that, in the latter half of the twentieth century, there was a discursive shift from education as a human right to education as a government investment, and that this shift refocussed public attention on both national standards and individual achievement. Through an examination of the structural changes to New Zealand education in the late 1980s and early 1990s, and the managerial discourses that accompanied them, the reasons why new technologies for measuring and monitoring learning and performance evolved are examined. But, in contrast to these accountability discourses, it is further argued that other professional assessment discourses were also competing for teachers’ attention at the time this investigation took place. A review of the international assessment research literature reveals evidence to support the view that, instead of increasing accountability mechanisms to raise standards, formative assessment strategies are critical in promoting learning and raising standards. With these findings in mind, 12 teachers at two Waikato primary schools and 20 teachers at 13 other New Zealand primary and intermediate schools participated in semi-structured interviews about their assessment practices. Three teachers were then each observed in their classroom environment for a week. Foucault’s analytics of power (namely discourse, disciplinary power, hierarchical observation, normalisation and the “examination”), are used to explain how and why, even though they valued formative assessment and child-centred pedagogies highly, many teachers predominantly employed summative assessment for accountability reasons. Due to increased managerial surveillance through the mechanisms of school and syndicate teamwork, appraisal, performance management and external audit, teachers have put themselves under pressure to account for their own practice. They have done this by finding ways to meet the documentation demands of the system, and in the process produced a summative shift in their assessment practice. Furthermore, although teachers were observed using formative assessment strategies known to improve learning, most did not identify these strategies as assessment and often understood the use of continuous summative assessment as formative. To address these findings, a case is made for professional development in assessment for school leaders and education officials as well as teachers. It is recommended that national and school policies of assessment, teacher appraisal and performance management be examined and revised to ensure that they produce predominantly formative rather than summative assessment practices.