Objective: This research examined the validity of criteria for diagnosing social phobia (SOC) and generalized anxiety disorder (GAD), where the "DSM-IV" criteria were modified to better identify toddlers who could have these disorders. Method: Diagnoses were made with a semistructured clinical interview that included child observations. Parents and caregivers completed child behavior, temperament, and socioemotional functioning questionnaires to test convergent and discriminant validity. Results: Of 72 children, 18 months to 5 years old, 19 met modified SOC criteria (8 met "DSM-IV" criteria SOC also), 29 met modified GAD criteria (5 met "DSM-IV" criteria GAD also), and 35 met no anxiety disorder criteria. Children with modified SOC were more likely than nonanxious children to display higher levels of anxiety symptoms and shyness/inhibition and to have anxious parents. Modified SOC did not relate to the nonanxiety constructs (cuddling, imaginary play, fine motor). Children with modified GAD did not consistently demonstrate higher levels of anxiety symptoms, did not have more anxious parents than nonanxious children, and did not have higher mean scores on the nonanxiety constructs. Conclusions: This research provides initial evidence supporting convergent and discriminant validity for the modified SOC criteria but not the modified GAD criteria. (Contains 2 tables.)