Abstract: | This study tested the utility of 29 maternal and familial characteristics for the purpose of prospectively identifying children who are at high risk for antisocial and delinquent outcomes. The family data were drawn from the archives of the Dunedin Multidisciplinary Health and Development Study. The study's design offers certain methodological advantages: the sample is a representative unselected birth cohort; the family measures were taken very early in childhood; information about the child's antisocial behaviour was collected from many different sources and at many different ages; a comparison group of children with other behaviour disorders was included, and it was possible to examine the influence of possible ‘confounding variables'. Three groups of 11-year-old children (antisocial (n = 50), other disorders (n = 37), and non-disordered (n = 220)) were compared on family variables. Nine family variables differentiated the antisocial children from the nondisordered children, the most important of which were parental disagreement about how to discipline the 5-year-old child, and many changes of the child's primary caretaker during childhood. In addition, among the children who were known to police by age 15, prospective family variables accounted for significant amounts of the variance in number of police contacts and age at first police contact. |