Background A youth justice diversion scheme designed to enhance health provision for young people with mental health and developmental problems as soon as they enter the youth justice system has been piloted in six areas of England. Aim As part of a wider evaluation of the first youth justice diversion scheme outside the USA, our aim here was to examine re‐offending. We sought to test the hypothesis that a specialised service for young people with mental health difficulties would be associated with reductions in re‐offending. In addition, we examined factors associated with the re‐offending that occurred. Methods Two hundred and eight young offenders with access to the diversion scheme and 200 without were compared in four geographical area pairings to allow for socio‐demographic contextual differences. Officially recorded re‐offending was ascertained for 15–30 months after study entry. We also tested characteristics associated with re‐offending among everyone entering the diversion scheme (n = 870). Results There was no statistically significant difference in re‐offending rates between the diversion and comparison samples, but those with access to diversion had significantly longer periods of desistance from offending than those who did not. In multivariate analysis, the only significant characteristic associated with re‐offending was history of previous offending. Conclusions Prevention of re‐offending is only one of the potentially beneficial outcomes of diversion of young people who are vulnerable because of mental health problems, but it is an important one. The advantage of longer survival without prevention of re‐offending suggests that future research should explore critical timings for these young people. The equivocal nature of the findings suggests that a randomised controlled trial would be justified. Copyright © 2014 John Wiley & Sons, Ltd.