The purpose of this preliminary study was to examine the validity of child, parent, and teacher reports of child social anxiety in predicting the child's responses to a social evaluative task. Children, parents, and teachers each completed a measure of social anxiety, as well as a measure that asked them to predict the child's anxiety during a behavioral approach task (BAT) of reading aloud in front of a video camera. Consistent with previous literature, analyses revealed poor agreement across informants. Importantly, children's reports predicted their own anxious feelings and, to some degree, their behaviors during the task, whereas other informants' reports did not. We recommend that further research examine the relative validity of multiple informants' reports in predicting children's real-life behavior.