We explored the utility of the Expressed Emotion (EE) construct with a community sample of young children (N = 91), studying the stability of EE scores over a 2-year period and the extent to which EE was associated with simultaneous and subsequent disruptive behavior. As part of a longitudinal study, families were assessed when the child was in preschool, 1st grade, and 3rd grade. Maternal EE was measured at preschool and 1st grade using the Five Minute Speech Sample and behavior ratings were obtained using the Child Behavior Checklist. At 3rd grade, the Diagnostic Interview Schedule for Children was used to determine DSM-IV diagnoses. The stability of the EE rating over the 2-year period from preschool to 1st grade was statistically significant although modest. At 1st grade, EE ratings were significantly related to the extent of externalizing behavior problems. Regression analyses that controlled for maternal stress levels determined that preschool EE ratings predicted classification of ADHD over 4 years later, at 3rd grade. The relationships between EE and child problem behavior were almost exclusively determined by the EE criticism dimension; the emotional overinvolvement dimension was not related to child behavior. The implications of these findings are discussed.