A new systemic approach to investigating child abuse deaths is proposed, drawing on the lessons learned in engineering. Investigations have traditionally taken the approach of concluding once faults in professional practice are identified. Solutions take the form of trying to control erratic practitioners: psychological pressure to achieve higher standards, increasing formalization and guidelines to reduce the scope for individual fallibility, and stricter management surveillance. The inquiry into the death of Victoria Climbie fits this model. However, thirty years of such inquiries have not led to the expected improvement in professional practice. Indeed, the Climbie report describes several agencies operating at a very low level, and failing to implement the most basic elements of good practice. A similar history of failure in engineering has led to the development of a systems approach. Human error is taken as the starting point, not the conclusion, and the investigation tries to understand why the mistake was made, by studying interacting factors in the practitioners, the resources available and the organizational context. The way this approach could be adopted in child protection work is outlined.