Background: The identification of the factors that influence the persistence of psychiatric disorder may assist practitioners to focus on young people who are particularly prone to poor outcomes, but population-based samples of sufficient size are rare. Method: This secondary analysis combined data from two large, population-based cross-sectional surveys in Great Britain (1999 and 2004) and their respective follow-ups (2002 and 2007), to study homotypic persistence among the 998 school-age children with psychiatric disorder at baseline. Psychiatric disorder was measured using the Development and Well-Being Assessment applying DSM-IV criteria. Factors relating to the child, family, and the severity and type of psychopathology at baseline were analysed using logistic regression. Results: Approximately 50% of children with at least one psychiatric disorder were assigned the same diagnostic grouping at 3-year follow-up. Persistent attention-deficit/hyperactivity disorder and anxiety were predicted by poor peer relationship scores. Persistent conduct disorder was predicted by intellectual disability, rented housing, large family size, poor family function and by severer baseline psychopathology scores. Conclusions: Homotypic persistence was predicted by different factors for different groups of psychiatric disorders. Experimental research in clinical samples should explore whether these factors also influence response to interventions.