Research Question: Does analysis of intimate partner violence (IPV) among Aboriginal and non-Aboriginal couples in the Northern Territory (NT), Australia reveal any predictable escalation in frequency or severity of harm over a four-year observation period? Data: We examined all 61,796 incidents of IPV recorded by the NT Police for 23,104 unique couples (‘dyads’), over the five year period from 1 January 2010 to 31 December 2014. For purposes of analysing changes over time in frequency and harm, we used standardized observation periods (generally four years) from first incident to end of observations. Methods: Each IPV incident was re-classified by crime type using the penal code of England and Wales, in order to measure the severity of harm in NT with the Cambridge Crime Harm Index (CHI). The CHI scores were used to test for patterns of concentration and escalation, based on the total days of recommended imprisonment for each offence type, summed across all offences of that type for the entire sample. Findings: The findings were sharply split between Aboriginal and white dyads. While there was no evidence of escalation in either frequency or severity of IPV 2 incidents in the white dyads, there was substantial evidence of escalation among Aboriginal offenders with three or more incidents in a four year period. Less than 2% of white offenders (2 of 111) had three or more incidents in four years, compared to 32.4% of Aboriginals (N = 105 out of 355 offenders). For those couples of both races known by police to have two or more incidents, there was a strong pattern of escalation in the frequency and seriousness of offending for up to 20 incidents over four years. While 66% of couples had desisted by year 3 with no further reports that year or the next, among the 34% of couples (N= 3,621) persisting into year 3 the probability of a new incident by year 4 was 99.9%. Similarly, the time between incidents for these repeaters declined with each new incident, indicating an increase in frequency. Severity of harm also rose with repeated incidents, from 0.6 of expected Cambridge CHI value per dyad among couples with 1 to 5 incidents over four years to 3.82 times higher than expected value per dyad among those couples observed to have 16-20 incidents over four years—six times more harm among couples (almost entirely Aboriginals) with the highest frequency of incidents than among couples with the lowest frequency. Conclusions: This targeting analysis confirms other research that shows no escalation in frequency or severity of domestic abuse among predominantly white European populations. Yet it also provides the first systematic test of the escalation hypothesis about IPV reported to police among Australian Aboriginal dyads. That evidence provides a strong basis in evidence for developing a two-track policy for policing IPV in Australian areas with substantial Indigenous populations. Track 1 would serve dyads (of either race) presenting for the first or second time, for whom a light touch may generally be sufficient. Yet any couple known to have had two or more prior offences could receive a far more intensive 3 strategic investment, including the testing of new strategies for prevention of escalation in harm or frequency of IPV. Yet because this pattern of escalation is found only in a minority of Aboriginal dyads, it is important to base policy on evidence-based targeting of dyads with prior occurrences rather than race. KEY WORDS: Intimate Partner Violence/Domestic Abuse/ Aboriginal Offenders and Victims/ Police/ Evidence-Based Targeting