Past research suggests that young children are incapable of reporting information about their own behavior problems. To test this, we examined the validity and the usefulness of children's self-reports in the E-Risk Study, a nationally representative birth cohort of 2,232 children. We used the Berkeley Puppet Interview to obtain children's self-reports of conduct problems when they were 5-years old and the Dominic-R when they were 7-years old. We also collected information about the children and their families by interviewing mothers, sending questionnaires to teachers, and rating examiners' observations during home visits. Results indicate that when children's self-reports are gathered with structured and developmentally appropriate instruments, they are shown to be valid measures: conduct problems reported by the children themselves were associated with known correlates including individual characteristics (e.g., IQ), related behaviors (e.g., hyperactivity), and family variables (e.g., economic disadvantages). Observed correlations closely matched effect sizes reported in the literature using adults' reports of children's behavioral problems. In addition, children's self-reports can be useful: both measures distinguished children meeting DSM-IV criteria for research diagnoses of conduct disorder. Children's reports also contributed unique information not provided by adults. For research and clinical purposes, young children's self-reports can be viewed as a valuable complement to adults' ratings and observational measures of children's behavior problems.