Alterations in cardiac shape and function, discerned by three-dimensional cardiac modeling, can provide important patient-specific diagnostic and prognostic information. However, the lack of standardized reference models for pediatric cardiac morphology is a barrier to their potential utility in pediatric heart diseases. In this study, we generated biventricular shape models from a retrospective analysis of 50 cardiac magnetic resonance imaging studies performed on pediatric patients from a single imaging center (ages 10–18 years) without known heart disease. We used principal component analysis to derive statistical modes of biventricular shape and function variation. The first principal component mode, which primarily reflected overall ventricular size, explained 20.2% of variation within the cohort. Values of this first mode, which corresponded to size and sphericity, were associated with biventricular volumes, body mass index, and diastolic blood pressure (all p < 0.05). Values of the second mode, which corresponded to size, sphericity, and function, were associated with biventricular volumes, ejection fractions, and heart rate (all p < 0.005). Comparisons of shape between this pediatric population and a healthy adult reference population revealed notable similarities and differences. This novel atlas provides a foundation that may be built upon with further data to better understand normal pediatric cardiac morphology and function and may serve as a valuable reference for pediatric heart disease states.