We welcome interest in our study of homicide in England and Wales. However, we disagree with Coid's assertion that the conclusions are illogical because the same social factors that were associated with the increase in homicides by the mentally ill up to the 1970s were present when those homicides declined. There are several possible reasons for decline in homicide by the mentally ill, including the availability of treatment. Coid's assertion that a fall in homicide due to better treatment must mean that the earlier rise was due to deteriorating mental health services is a similar oversimplification.

There has been no change in the law regarding diminished responsibility since 1957. Coid's explanation that the decline in homicide by the mentally ill since the late 1970s was due to a change in the threshold for the verdict of diminished responsibility is not supported by any data. Moreover, a change in threshold for diminished responsibility would not explain the decline in the verdicts of ‘not guilty due to mental illness’, ‘permanently unfit for trial’ and ‘infanticide’. We also defend the use of legal outcomes to define cases. Given the careful attention paid to homicide matters by the courts, their verdicts are likely to be reasonably sensitive and highly specific. ^{Reference Nielssen and Large1}

Vinkers et al report 8 years of data from The Netherlands, without showing that rates of homicide by the mentally ill have declined over a longer period. However, a lack of a decline in The Netherlands might not be unexpected, as we have found that 40% of homicides in psychotic illness occur before treatment, ^{Reference Nielssen and Large1} that delay in the initial treatment of schizophrenia is associated with a greater proportion of homicides during the first episode of psychosis ^{Reference Large and Nielssen2} and that jurisdictions with mental health laws that require a patient to be dangerous before they can receive involuntary psychiatric treatment, such as The Netherlands, have longer delays in the treatment of early psychosis. ^{Reference Large, Nielssen, Ryan and Hayes3}

We look forward to a challenge to our findings based on data rather than opinion and speculation.