This study sought to assess the impact of the pre-sentence Magistrates Early Referral Into Treatment (MERIT) diversion program in New South Wales, Australia on offending in the 12 months following exposure to the intervention. The comparative design involved an experimental group of 1017 defendants who exited the MERIT program in 2008 and a comparison group of 1017 offenders identified as drug misusers following completion of a Corrective Services’ risk and needs assessment and sentencing in a non-MERIT New South Wales Local Court during this period. Both groups were selected from a larger pool of cases (N = 3319) that were found to be significantly different on a number of key variables. Propensity score matching was used to deal with this non-equivalence. The outcome measures were: the rate, volume and seriousness of known re-offending. Recidivism risk factors were also examined for MERIT participants. There was no association between exposure to MERIT and reduced rates of reconviction at 12 months and group status did not make a significant contribution to the prediction of reconviction, when controlling for the influence of penalties imposed. The MERIT group had committed a larger number of offences during this period (including when adjustments were made for time at reduced risk due to imprisonment), but there were no differences in relation to changes in offence seriousness. Among MERIT participants, the factor with the largest effect on risk of recidivism was offence type. Completing the program had a significant protective effect against recidivism, while the number of prior court convictions increased this risk. These findings shed light on possible reforms to MERIT aimed at maximising the program’s crime prevention impact.