This review focuses on ways in which epidemiological research can inform mental health service development and clinical practice. Data from epidemiological studies can provide cross-sectional and secular estimates of the prevalence of psychopathology to support rational service development. Epidemiological surveys have difficulties in finding large enough samples of children with rare disorders, although these disorders are often severely debilitating and require extensive service input. Systematic surveillance provides a rigorous method for studying rare disorders and events. Only a minority of children with impairing psychopathology reach mental health services, although a larger proportion have mental health related contacts with other services. The gap in provision is such that an expansion of mental health services is unlikely to reach all children who could benefit, suggesting that mental health professionals need to develop innovative strategies to increase the number of children seen and the effectiveness of interventions that they receive. Training and supervision of non-mental-health professionals working with children in the identification and management of mental health problems is also extremely important. Most studies suggest that the children with the severest problems are getting to specialist mental health services, and service contact is more likely if important adults can perceive the child's difficulty or find it to be burdensome. The latter suggests that education of key adults would improve detection if services had the capacity to cope. Studies consistently suggest that the region in which the child lives affects the likelihood of service contact, but studies of other characteristics predicting service contact are so contradictory that studies should only be (cautiously) applied to similar populations to assess which types of children may currently be falling through gaps in service provision. Academics are beginning to explore the use of structured measures developed for epidemiological studies in clinical assessment and outcome monitoring.