The present study examines the construct validity of separation anxiety disorder (SAD), social phobia (SoP), panic disorder (PD), and generalized anxiety disorder (GAD) in a clinical sample of children. Participants were 174 children, 6 to 17 years old (94 boys) who had undergone a diagnostic evaluation at a university hospital based clinic. Parent and child ratings of symptom severity were assessed using the Multidimensional Anxiety Scale for Children (MASC). Diagnostician ratings were obtained from the Anxiety Disorders Interview Schedule for Children and Parents (ADIS: C/P). Discriminant and convergent validity were assessed using confirmatory factor analytic techniques to test a multitrait-multimethod model. Confirmatory factor analyses supported the current classification of these child anxiety disorders. The disorders demonstrated statistical independence from each other (discriminant validity of traits), the model fit better when the anxiety syndromes were specified than when no specific syndromes were specified (convergent validity), and the methods of assessment yielded distinguishable, unique types of information about child anxiety (discriminant validity of methods). Using a multi-informant approach, these findings support the distinctions between childhood anxiety disorders as delineated in the current classification system, suggesting that disagreement between informants in psychometric studies of child anxiety measures is not due to poor construct validity of these anxiety syndromes. (Contains 3 tables and 1 figure.)