Following a pre-registered study design, we estimated the effect of police force-wide domestic abuse training on arrests for the new crime of ‘controlling or coercive behaviour’. Using data on monthly counts of arrest for controlling or coercive behaviour in 33 police forces, we performed a negative binomial difference-in-difference analysis and capitalised on differences in intervention timing to undertake an event study. Training was associated with a 41% increase in arrest for controlling or coercive behaviour for trained forces (IRR 1.413, 95% CI 1.235–1.617) and the increase in arrests was consistent with the timing of the training. This study provides evidence that training entire police forces to understand the dynamics of domestic abuse, including the new offence of coercive control, is effective in increasing the rate of arrest for coercive control. However, the number of coercive control arrests as a proportion of total domestic abuse arrests remains miniscule.