Introduction: Various T-wave characteristics have been shown to be predictive of ventricular diastolic dysfunction and sudden cardiac death in adults. However, specific T-wave morphologies and their predictive potential in pediatric patients is unknown. Hypothesis: Machine learning derived T-wave characteristics will be different between children with pulmonary arterial hypertension (PAH) and healthy controls. Furthermore, T-wave characteristics will be predictive of clinical functional worsening in PAH patients. Methods: Principal component analysis (PCA) was applied to ECGs collected for children with PAH (n=155) at the time of diagnostic catheterization and matched healthy controls (n=47) to describe basic T-wave characteristics. Patients (<18 y/o) with PAH were defined as a mean pulmonary pressure >25 mmHg in the absence of obstructive left heart disease. Interpreted T-wave features were then compared between PAH and control groups. Identified T-wave measurements underwent survival analysis using standardized PAH functional class worsening as a clinical endpoint. Results: Identified T-wave principal components were 1) T-wave height and 2) early or late incidence of T-wave peak (FIGURE- 1A). The second principal component was associated with PAH diagnostic specificity (94%). Time from T-wave peak to T-wave end (TpTe) as measured in leads V4 to V6 was significantly higher in children with PAH (all P < 0.001) (FIGURE - 1B). V4 derived TpTe value > 132 ms and V5 derived TpTe value > 135 ms were associated with the higher probability of clinical functional worsening (P = 0.018 and P = 0.002, respectively) (FIGURE - 1C). Conclusions: T-wave features derived from standard ECG are 1) different between children with PAH and their healthy peers and 2) predictive of clinical functional worsening.