When a convicted or alleged child sex offender is living, or having contact, with his own children or stepchildren, the obvious worry is that these children are victims or will become victims of sexual abuse. One way of determining the risk of this occurring is for the convicted or alleged offender to undergo a forensic sex offender risk assessment. In this article I raise questions regarding the usefulness of sex offender risk assessments within the statutory child protection context. Most importantly, I ask whether static and dynamic risk assessment instruments can accurately predict the risk an alleged or convicted sex offender poses to his own children. I conclude that ‘high’, ‘moderate’, and ‘low’ risk outcomes of forensic sex offender risk assessments in the child protection context are unreliable and can result in error, and explain that these errors have consequences that, within the child protection context, have consequences that can be dangerous to children.