This study addresses the prediction that dropout from a UK specialized residential treatment program for adolescent sexual abusers can be determined from pre-treatment variables. Participants were 49 adolescents aged 12-16 years, who had sexually abused children, peers/adults or both. Of the variables examined, 25 showed a significant association with treatment dropout. A scale, consisting of 20 items, was designed to predict treatment dropout. As a measure it showed internal consistency (alpha = 0.84) and predictive validity. Treatment dropout was linked to a greater risk of recidivism: offences of a general; violent; and combined violent or sexual nature. Missing data confounded the overall small sample size; therefore, a brief checklist of factors associated with dropout was produced as a guide for treatment managers and clinicians. (Contains 2 tables and 2 figures.)