This review evaluates the clinical utility of tools for systematic behavioral observation in different settings for children and adolescents with ADHD. A comprehensive search yielded 135 relevant results since 1990. Observations from naturalistic settings were grouped into observations of classroom behavior (n = 58) and of social interactions (n = 25). Laboratory observations were subdivided into four contexts: independent play (n = 9), test session (n = 27), parent interaction (n = 11), and peer interaction (n = 5). Clinically relevant aspects of reliability and validity of employed instruments are reviewed. The results confirm the usefulness of systematic observations. However, no procedure can be recommended as a stand-alone diagnostic method. Psychometric properties are often unsatisfactory, which reduces the validity of observational methods, particularly for measuring treatment outcome. Further efforts are needed to improve the specificity of observational methods with regard to the discrimination of comorbidities and other disorders.