Objective: To evaluate the impact of changing the response labels of a teacher rating scale in a population-based study of ADHD. Method: For parents, the Diagnostic Interview Schedule for Children, which asks whether each of 18 symptoms occurred "often" in the past year, was used. For teachers, most scales use a 4-point scale, with "often" and "very often" as the two highest categories. To make the two instruments more compatible, the authors created a teacher rating scale with "often" as the highest category. Teachers rated more than 6,000 elementary school students using this instrument. Some teachers completed ratings using both questionnaire formats. Results: The scale's internal consistency, convergent validity, and test-retest reliability are excellent. The factor structure is consistent with established "DSM-IV"-based instruments. Response category choices influence the proportion of children exceeding symptom and impairment thresholds. Conclusion: Altering the response scale does not degrade the psychometric properties of the instrument but makes important differences in the sample. (Contains 6 tables and 4 figures.)