BACKGROUND: Health service reconfigurations are of international interest but remain poorly understood. This article focuses on the use of evidence by senior managerial decision-makers involved in the reconfiguration of stroke services in London 2008-2012. Recent work comparing stroke service reconfiguration in London and Manchester emphasises the ability of senior managerial decision-makers in London to 'hold the line' in the crucial early phases of the stroke reconfiguration programme. In this article, we explore in detail how these decision-makers 'held the line' and ask what the broader power implications of doing so are for the interaction between evidence, health policy and system redesign. METHODS: The research combined semi-structured interviews (n = 20) and documentary analysis of historically relevant policy papers and contemporary stroke reconfiguration documentation published by NHS London and other interested parties (n = 125). We applied a critical interpretive and reflexive approach to the analysis of the data. RESULTS: We identified two forms of power which senior managerial decision-makers drew upon in order to 'hold the line'. Firstly, discursive power, which through an emphasis on evidence, better patient outcomes, professional support and clinical credibility alongside a tightly managed consultation process, helped to set an agenda that was broadly receptive to the overall decision to change stroke services in the capital in a radical way. Secondly, once the essential parameters of the decision to change services had been agreed, senior managerial decision-makers 'held the line' through hierarchical New Public Management style power to minimise the traditional pressures to de-radicalise the reconfiguration through 'top down' decision-making. CONCLUSIONS: We problematise the concept of 'holding the line' and explore the power implications of such managerial approaches in the early phases of health service reconfiguration. We highlight the importance of evidence for senior managerial decision-makers in agenda setting and the limitations of clinical research findings in guiding politically sensitive policy decisions which impact upon regional healthcare systems.