Objective: This thesis set out to investigate whether there were differences between juveniles who had either sexually offended against children (child molesters) or against adult women/female peers (sexual assaulters), and between both these groups and non-sexual violent offenders and property offenders. The groups were compared across a range of demographic, developmental, family, educational, offence, and psychological variables, including IQ, attribution of criminal behaviour, masculine identity, perception of parenting style, relationship with parents, and attitudes toward rape and rape victims. Method: Four groups of 27 male British juveniles who admitted their offending were compared, matched for age, ethnicity, and socioeconomic status. Many of the participants had been referred to Glenthorne Youth Treatment Centre, and had committed the most serious types of sexual and violent offences. Background data were collated from case-notes and other archival sources using a structured proforma. Participants were also seen by the researcher for administration of a battery of psychometric tests and questionnaires. Results: A number of between-group differences were found, especially in the child molesters, who were found to have significantly more social, developmental, and learning problems; to be less delinquent and antisocial; more extensively sexually victimized; and less violent in their sexual offending. However, a minority had been sexually violent. The sexual assaulters were more likely to be dissatisfied with parental care and relationships, especially with father; have a hypermasculine sexual identity; and to hold calloused sexual attitudes, and rape-supportive beliefs and attitudes. The gender-socialization theory of male sexual aggression toward women was supported. A subgroup of extremely delinquent, violent juvenile sex offenders was also identified, composed of child molesters and sexual assaulters. Conclusions: The differences found between sexual assaulters and child molesters, and the existence of a mixed-group of extremely antisocial juvenile sex offenders, have implications for the treatment and classification of juvenile sex offenders. Generic intervention programmes for young offenders should adopt behavioural and thinking skill-development approaches to reduce the risk of criminal behaviour, including sexual crime. Sex-offencespecific treatment should be targeted more closely at the minority of high-risk, sexually deviant young offenders. Typologies of juvenile sex offenders also need to be revisited