Background: It is an increasing priority for clinicians and lawyers to define the evidence base in relation to child abuse. Due to the paucity of evidence, we have conducted a series of systematic reviews to define the evidence base underpinning the identification and investigation of cases of suspected physical child abuse. Although Evidence Based Medicine is traditionally dominated by the randomized controlled trial as the gold standard, this research field can only be addressed through observational studies. Objective: To outline the methodology used to address the questions: how do you distinguish inflicted from non-inflicted brain injury? What is the optimal imaging strategy to explain these injuries? Methods: An all language literature search was performed across 20 databases, information sources and relevant websites of published or unpublished studies for the years 1970 2008. In addition, a range of snowballing techniques were used to increase the sensitivity of the search. The search retrieved in total 7802 references which were imported into bibliographic software. The titles were scanned for duplicates and relevancy, 5861 references were sent to principle investigator for the second scan based on inclusion criteria, of which 305 studies were reviewed. Results: We have identified, by multilevel logistic regression analysis, key clinical features that predict inflicted brain injury and clarified the optimal investigation strategy. The findings informed TheNICE guidance When to suspect child maltreatment http://guidance.nice.org.uk/CG89 and the recent joint Royal College of Paediatrics and Child health and Royal College of Radiology Standard for diagnostic imaging of suspected non-accidental injury in children. Findings are disseminated by peer review publications, a dedicated website www.core-info.cf.ac.uk and leaflets for other childcare professionals. Conclusion: This review challenges long held views of paediatricians and provides an evidence base for child protection professionals. It has identified research priorities where evidence is lacking.