Aim To report the retinal signs that distinguish abusive head trauma (AHT) from non-abusive head trauma (nAHT). Methods A systematic review of literature, 1950-2009, was conducted with standardised critical appraisal. Inclusion criteria were a strict confirmation of the aetiology, children aged o11 years and details of an examination conducted by an ophthalmologist. Post mortem data, organic disease of eye, and inadequate examinations were excluded. A multivariate logistic regression analysis was conducted to determine odds ratios (OR) and probabilities for AHT. Results Of the 62 included studies, 13 provided prevalence data (998 children, 504 AHT). Overall, retinal haemorrhages (RH) were found in 78% of AHT vs 5% of nAHT. In a child with head trauma and RH, the OR that this is AHT is 14.7 (95% confidence intervals 6.39, 33.62) and the probability of abuse is 91%. Where recorded, RH were bilateral in 83% of AHT compared with 8.3% in nAHT. RH were numerous in AHT, and few in nAHT located in the posterior pole, with only 10% extending to periphery. True prevalence of additional features, for example, retinal folds, could not be determined. Conclusions Our systematic review confirms that although certain patterns of RH were far commoner in AHT, namely large numbers of RH in both the eyes, present in all layers of the retina, and extension into the periphery, there was no retinal sign that was unique to abusive injury. RH are rare in accidental trauma and, when present, are predominantly unilateral, few in number and in the posterior pole.