This study evaluated the accuracy of teachers' ratings and examined whether these ratings are influenced by halo effects. One hundred thirty-nine elementary school teachers viewed videotapes of what they believed were children in regular fourth-grade classrooms. In fact, the children were actors who followed prepared scripts that depicted a child engaging in behaviors characteristic of an attention-deficit hyperactivity disorder (ADHD), an oppositional defiant disorder or a normal youngster. The findings provide support for a bias that was unidirectional in nature. Specifically, teachers rated hyperactive behaviors accurately when the child behaved like an ADHD youngster. However, ratings of hyperactivity and of ADHD symptomatic behaviors were spuriously inflated when behaviors associated with oppositional defiant disorder occurred. In contrast, teachers rated oppositional and conduct problem behaviors accurately, regardless of the presence of hyperactive behaviors. The implications of these findings regarding diagnostic practices and rating scale formats are discussed.