Objective: To identify ADHD symptoms that are most highly predictive of ADHD diagnostic status and compare them against Diagnostic and Statistical Manual of Mental Disorders (4th ed.; DSM-IV) algorithms in predictions of functional impairment. Method: Parent and teacher ratings of ADHD were obtained from an ethnically diverse (46% non-White) sample of 151 five- to ten-year-old children (27% female) with (n = 76) and without (n = 75) DSM-IV ADHD. We calculated total predictive values to estimate how ratings of each ADHD symptom predicted ADHD diagnostic status based on a structured parent diagnostic interview. Optimal symptom thresholds (i.e., not at all, just a little, pretty much, very much) for predicting ADHD caseness differed for inattention and hyperactivity and parents versus teachers. Algorithms consisting of highly predictive symptoms were then created and compared against DSM-IV-based algorithms to predict independent measures of functional impairment. Results: Several empirically derived symptom algorithms were more strongly associated with functional impairment than DSM-IV-based algorithms. Conclusion: These preliminary findings suggest that alternative methods to aggregating ADHD symptoms may improve predictions of impairment. (J. of Att. Dis. 2014; XX(X) 1-XX)