How do judges assess witness credibility? This question is particularly important in cases where child victims testify. These cases are usually of a sensitive nature (e.g., involving allegations of abuse) where there are no other testimonial or material evidence to corroborate the account of child victims. To better understand what actually happens when judges preside over trials about sexual interference, and how, in actual courtrooms, they use demeanour to assess the credibility of child victims, we conducted a qualitative thematic analysis of real courts judgements (n = 44). The results highlighted that when assessing credibility in actual courtrooms, judges make a variety of inferences from the demeanour of child victims who testify live, with striking differences (and similarities) when defendants are found not guilty than when they are found guilty. The results of our descriptive study also provided unique insights about (correct and incorrect) beliefs judges hold about child witnesses, and how, if children fail to display behaviour they are expected to display, or if they display behaviour they are not expected to display, child victims could face difficulties at several stages of the judicial process. We discuss the results based on the literature on nonverbal behaviour and child witnesses, explaining their scope for scholars and legal practitioners.