In cases of suspected child maltreatment, the caseworkers’ evaluations of the harm and risk to the child are vital in determining if children are being abused and ultimately whether services are provided to the family. These evaluations are dependent on information caseworkers are able to uncover during their investigation, but may not reflect the experiences of the child. Using data from the National Survey of Child and Adolescent Well-Being (NSCAW-I), this study first compares how consistent children’s claims of physical victimization are with caseworkers’ assessments of harm, severity of risk, and whether there is physical abuse occurring. Second, we examine whether any discrepancies are influenced by the demographic characteristics of the child. Based on cross-tabulations, mean tests of difference, and one-way ANOVA, we find a high degree of overlap in children’s reports of violence with caseworkers’ reports of harm, the potential for risk, and the presence of physical abuse. However, there are also important differences. Among children who reported acts of physical violence occurred "lots of times," 23% of the cases were viewed by the caseworker as causing "no harm" to the child and over 60% were not regarded as physical abuse. The children’s age and sex are both significant predictors of discrepancies, with more discrepancies between caseworkers and young children or teens as well as boys. Implications of this study are that additional training is needed to help caseworkers build rapport with the children. We also suggest that Child Protective Services (CPS) should implement alternative ways for children to report their experiences other than face-to-face interviews with caseworkers.