Research Question: Can integrated case management by a multi-agency partnership of the relations between offenders and victims with repeated incidents of intimate partner violence (IPV) reduce the frequency or severity of harm from that violence? Data: Three batches of 60 IPV dyads were enrolled in a trial, with data collected on services delivered to them and police records for 2 years before and 2 years after random assignment to treatment and control groups. Methods: The study measured the delivery of all three elements of treatment offered: (1) victim support through Berkshire Women’s Aid, (2) one-to-one perpetrator counselling through motivational interviewing techniques and (3) follow-up visits to the home addresses of perpetrators and victims. The outcomes for each couple in severity of harm were compared in a before-after, difference-of-differences analysis of Cambridge Crime Harm Index scores. After-only frequency of non-criminal domestic conflict events was also compared. Findings: Delivery of programme elements was highly variable, but more intense in the treatment group than in control, especially in terms of police visits to offenders (T = 60%, C = zero). Mean difference between 24 months of post-random assignment and the 24 months baseline period for C cases was an increase of 4.15 Cambridge Crime Harm Index (CHI) prison days, while T cases had a mean change of 8.85 fewer CHI days in prison in post-assignment than in baseline. This difference was significant with outliers removed, but not with two control group baseline cases included. There was also a substantially higher rate of frequency of non-crime events in the 24 months after random assignment in T (112) than in C (85). Conclusions: The overall effect of the programme appeared to have been beneficial, as measured by the Crime Harm Index. The evidence cannot specify how much of that benefit was caused by the more consistent police visits to offenders versus other elements of the programme for both victims and offenders.