BACKGROUND: The rise in homicides by those with serious mental illness is of concern, although this increase may not be continuing. AIMS: To examine rates of mental illness among homicide perpetrators. METHOD: A national consecutive case series of homicide perpetrators in England and Wales from 1997 to 2006. Rates of mental disorder were based on data from psychiatric reports, contact with psychiatric services, diminished responsibility verdict and hospital disposal. RESULTS: Of the 5884 homicides notified to the National Confidential Inquiry into Suicide and Homicide by People with Mental Illness between 1997 and 2006, the number of homicide perpetrators with schizophrenia increased at a rate of 4% per year, those with psychotic symptoms at the time of the offence increased by 6% per year. The number of verdicts of diminished responsibility decreased but no change was found in the number of perpetrators receiving a hospital order disposal. The likeliest explanation for the rise in homicide by people with psychosis is the misuse of drugs and/or alcohol, which our data show increased at a similar magnitude to homicides by those with psychotic symptoms. However, we are unable to demonstrate a causal association. Although the Poisson regression provides evidence of an upward trend in homicide by people with serious mental illness between 1997 and 2006, the number of homicides fell in the final 2 years of data collection, so these findings should be treated with caution. CONCLUSIONS: There appears to be a concomitant increase in drug misuse over the period, which may account for this rise in homicide. However, an increase in the number of people in contact with mental health services may suggest that access to mental health services is improving. Previous studies have used court verdicts such as diminished responsibility as a proxy measure of mental disorder. Our data indicate that this does not reflect accurately the prevalence of mental disorder in this population.