Background: Most studies on the predictors and effectiveness of community treatment orders (CTOs) are restricted to health-related variables and do not consider forensic contacts as established by criminal justice databases or predictors from birth. We used linked administrative health and criminal justice data for a birth cohort in Queensland, Australia to investigate the characteristics and outcomes of people placed on CTOs. Methods: CTOs were identified from administrative data for hospital admissions and community mental health service contacts for a population cohort of 45,141 individuals born in Queensland in 1990. These data were linked with administrative court records, with individuals followed up to age 23/24 years. Logistic regression analyses were used to examine characteristics associated with CTO placement and Tobit regression analyses to examine factors predicting health and criminal justice outcomes in the following year. Results: There were 211 CTO cases by age 23/24 years, for whom it was possible to identify 413 controls on voluntary treatment. Non-affective psychoses [F20-F29] were the strongest predictors of CTO placement (ORadj = 4.07, 2.77–5.99) followed by a court appearance (ORadj = 1.99, 1.28–3.09). CTOs were associated with greater, not lower, subsequent psychiatric hospital admissions, inpatient bed-days and community mental health service contacts, although on sensitivity analyses psychiatric hospital admissions were the same as voluntary controls. CTOs were not associated with more subsequent court appearances despite higher rates of offending before CTO placement. Conclusions: Both clinical and forensic variables can determine CTO placement and, on adjustment for these covariates, CTOs were not associated with reductions in psychiatric hospital admission, time spent as an inpatient, or subsequent court appearances. The latter finding might mean that CTOs reduce the risk of offending to that of voluntary controls.