The early detection of children with attention deficit hyperactivity disorder (ADHD) by teachers can contribute to the prevention of secondary disorders in a child and this can have serious implications for the child's overall development. The aims of the present study were to examine: (1) the validity of the original assessment of the teachers in detecting school-age children nominated as children with ADHD by their teachers; and (2) the factors that influence teachers' evaluations on a rating scale assessing ADHD, such as child's gender, the parents' level of education, the child's social and academic behaviour in the classroom, the teacher-student relationship, as well as the teachers' cooperation with the parents. Teachers were asked to fill in a Greek standardised rating scale for the detection of children with ADHD and a student behaviour questionnaire for 420 primary-school students nominated as children with ADHD by teachers. The study's findings showed that teachers detect far more students with ADHD than the number expected from the norms based on the standardised ADHD rating scale. Teachers not only nominated more boys than girls as students with ADHD, but also made more accurate identifications for girls than for boys, particularly in the later primary-school grades. Children's age or parental educational background did not influence the teachers' initial nominations or their later ratings. Multiple regression analyses indicated that the teachers' beliefs about a child's peer relations and the quality of teacher-child relationships predicted teachers' ratings on the ADHD scale more than other factors did. (Contains 4 tables and 1 figure.)