Introduction Quality data are important for health resource allocation and the design, implementation and evaluation of injury prevention programmes and policies. Such data are seldom available in resource poor settings. Although there have been successful attempts to systematically collect mortality data, the introduction and sustainability of hospital-based injury surveillance has been more challenging.

Materials and methods This study describes the implementation of two hospital-based injury surveillance pilots in Africa by International Physicians for the Prevention of Nuclear War, and a South African research consortium. It draws on the findings of process evaluations conducted during implementation. The examples of Zambias Lusaka University Hospital and South Africa's GF Jooste Hospital in Cape Town reveal similar recurrent themes.

Results The utility of the injury data was apparent to various stakeholder groups, and although both projects met with initial success, neither was sustainable. Key weaknesses invariably related to a lack of institutionalisation and local investment. For example, there was no centralised oversight to coordinate the projects in the absence of the implementing research agencies, new staff were not adequately trained in data collection and, as there was no formal inclusion in hospital protocols, data collection was regarded as research and conducted ad hoc.

Conclusion The findings have implications for future attempts to introduce routine injury surveillance in resource poor settings, and indicate a role for multilateral agencies in developing and promoting implementation guidelines for adoption at a national level, as well as investment by health authorities in quality data collection, analysis and reporting.