Unintentional injuries (accidents) are a global child health problem. Many child injury prevention measures are proven to be effective, yet government and community focus on prevention waxes and wanes through time and across locations. Within New Zealand some measures, such as child car seats, are mandated and enforced while the provision of other equally effective strategies, such as the enforcement of swimming pool fencing regulation, appears inconsistent. This research set out to discover what influences the government’s adoption of injury prevention policies and programmes. The research analysed interview and case study data using a Foucauldian understanding of everyday practice underpinned by the analysis of interview and case study data, while also applying critical and grounded theory and public policy research. Foucault’s concept of governmentality assisted in the exploration of government actions. The findings demonstrate support for improving child safety from government employees at various levels of responsibility across many agencies. Such support has been provided both with and without endorsement by political decision makers. Positivist, quantitative research is the foundation of injury prevention science, but at times can have perverse effects, especially if work to count and reduce injury events is construed as an effort to displace valued childhood experiences. Advocacy and lobbying for child injury prevention are acknowledged by those active in injury prevention as essential activities but are not well explored or researched within New Zealand. Collaboration has long been recommended as best practice for child injury prevention, because it enables wider distribution of messages and better use of resources. This research identified organisational cascades, where backbone organisations provided resources to other organisations, so they could also act as backbone organisations and support collaborative ventures. Collaboration can be counterproductive however, when child safety practitioners and advocates develop strongly coherent identities and reduce their communication with other groups. Such behaviour risks safety groups being unaware of emergent discourse that undermines injury prevention measures and sets them up to be negatively stereotyped and marginalised from decision making. This research highlights how success at preventing child injuries is contingent upon both positivist research and the presence of a widely accepted safety culture, where the use of safety equipment and safe practices are promoted by everyone as ‘the way we do things’. New Zealand’s child injury effort has been mostly effective, and injuries are reducing in number. Despite this, there is a risk that gains in child safety might be lost, should there be insufficient recognition of the factors that have been important for these improvements to have occurred. There is also the possibility successful initiatives might prompt their premature demise by fostering an impression that government’s support for unintentional child injury prevention is no longer justified. This research concludes with recommendations for injury prevention practitioners and researchers.