This paper examines the multi-agency identification of risk and harm in extra-familial Child Exploitation (CE). It explores several data prediction methods to effectively target and prevent harm. It provides a taxonomy analysis of repeat-victimisation and cumulative victim harm. It also examines the relationship between age and harm and finally, the conditional probability of repeat victimisation in exploited children. Research Question Are the most harmed exploited children referred to Multi-Agency Child Exploitation (MACE) Panels and what other methods exist to identify and prevent high harm in child exploitation? Methods This is a descriptive quantitative statistical analysis using a whole population of children in a Northern English county, aged between 10-17 who were recorded in police data as either victims, offenders or MACE referrals between January 2017 and June 2018, encompassing 12,457 children. It utilises an 18-month study window, an 18-month follow up and 18-month prior time censored period using data between 2015 and 2019 inclusive. This data identifies CE victims using CE flagging and additionally, offence classification with familial abuse and familial exploitation cases removed. It identifies repeat-victims and those children referred to MACE for tailored multi-agency intervention. Application of Sherman et al.’s (2016) Cambridge Crime Harm Index (CCHI) provided an analysis of harm in victimisation and offending. Findings 90.7% of exploited repeat-victims were not referred to MACE and there was no significant difference in the harm they sustained in the 18-months following a repeat victimisation compared to exploited children subject to MACE. The most harmed CE victim (Victim-CCHI 15,330) in the 18-month study window was not referred to MACE, nor was the highest frequency CE victim within 18-months (31 victimisations). Exploited victims, victim-offenders and MACE children are re-victimised at a significantly higher rate than other children. 73.4% of exploited victims will not suffer a repeat victimisation of any kind. 42% of exploited repeat-victims will have a third victimisation given a second and this will attract additional mean victim harm of 464 (For comparison, Penetrative Sexual activity with a girl under 16 by an Offender 18 or over has a CCHI harm score of 365). Conclusion Whilst MACE provides a forum to share multi-agency information, it only does so for 9.3% of exploited repeat-victims. This has implications for the role and focus of MACE. Whilst several quantitative methods were explored to predict harm in CE, this research favours the use of conditional probability and harm association. By using this method, 90.7% of missed repeat victims become visible to professionals. This is essential in providing the opportunity to minimise the risk of further victimisation and increased harm that 42% of this group will have within an 18-month period. This research provides a predictive and evidence-based framework to identify exploited children at risk of further harm and victimisation.