Our aim was to determine the overall rates of general and sexual re-offending of youth (i.e. aged under 18 at the time of offence) charged with sexual offences, ranging from indecent exposure to rape, over 6.5 years in South Australia and whose cases were finalised in court, by conference and by formal caution (N = 365). Controlling for previous offending, we examined if re-offending varied by site of finalisation or by referral to Mary Street, a specialist treatment program. Follow-up times ranged from six to 84 months. We applied a parametric form of survival analysis by fitting the Weibull 'mixture model' to the Kaplan-Meier cumulative distribution of failure times (time to re-offend). Covariates, such as prior offending or referral to Mary Street, were introduced to test for differences in survival rates, immune proportions or both between groups. By the cut-off date, 54% of youth had been charged with new non-sexual offences but only 9% with new sexual offences. Court youth had a higher rate of re-offending than conference youth, but these differences were largely explained by prior offending. For the subgroup with no previous offending, however, a significantly slower rate of re-offending was observed for conference youth and for those who were referred to Mary Street. We were able to control for the main effect of prior offending, but complex interactions between co-variates such as offence types, early admissions to offending and legal and therapeutic responses could not be disentangled in our small sample, and we could not explore factors linked specifically to sexual re-offending. Future research should examine the enmeshment of key factors that set youth on differing legal pathways; this phenomenon affects studies comparing conference and court outcomes, whether in naturalistic settings or in randomised field experiments. © The Author(s) 2013 Reprints and permissions: sagepub.co.uk/journalsPermissions.nav.