A representative sample of 328 children with hemiplegia was assessed by behavioural screening questionnaires completed by parents and teachers; just under half were also individually assessed. Four years later, 90% of the sample were followed up by questionnaire. There was substantial continuity across time for most measures of psychopathology. Depending on the measures used, around 70% of children who were psychiatric "cases" initially were still cases 4 years later. In addition, around 30% of children who were not psychiatric cases initially had become cases 4 years later. In the pre-school years, externalising symptoms were predictive of later conduct and hyperactivity problems, whereas emotional symptoms were not predictive of later problems. In the school years, hyperactivity was particularly predictive of continuing psychiatric problems. Neurological, cognitive, demographic, and family factors did not add much predictive power once allowance had been made for the severity and type of the initial psychiatric problems. Since the psychiatric complications of childhood hemiplegia are persistent as well as common, they warrant increased clinical and research attention. All children with hemiplegia deserve careful monitoring for psychiatric problems, with a view to early intervention or the deployment of preventative approaches.