Although major guidelines in the field and current diagnostic criteria clearly demand an assessment of children’s attention deficit/hyperactivity disorder (ADHD) symptoms at school, few studies address the fundamental question of which is the best approach for clinicians to get this information from teachers. Three screening strategies for ADHD were applied to teachers of 247 third grade students. They were asked (1) an overt question about potential cases of ADHD in their classroom; (2) to complete a broad-band questionnaire assessing common child mental health problems; (3) to rate ADHD-specific symptoms in a narrow-band questionnaire. Based on the overt question, teachers identified one in five students (21.1 %) as having ADHD; 28 cases (11.3 %) were identified using standard cut-offs for the narrow-band, and 13 (5.3 %) using a standard threshold for the sub-scale of hyperactivity from the broad-band questionnaire. Agreement among strategies was low (k = 0.28). A subsample of students, clinically assessed to confirm screenings, showed modest agreement with final diagnosis. The narrow-band questionnaire had the best diagnostic performance. Multivariate analysis indicated that the presence of a comorbid externalizing disorder was the only variable associated with teachers’ ascertainment of ADHD caseness or non-caseness. Choice of screening strategy significantly affects how teachers report on ADHD symptoms at school. The halo effect of externalizing behaviors impacts the correct identification of true cases of ADHD in the school setting. Clinicians can rely on narrow-band instruments like the SNAP-IV to get information on ADHD symptoms at school from teachers.