The native aorta in new-borns with hypoplastic left heart syndrome (HLHS) is often small or malformed. It requires surgical reconstruction and enlargement to enable systemic blood flow from the single right ventricle. However, morbidity and mortality remain high after surgery. The shape and size of the reconstruction are important factors affecting the hemodynamics in the aortic arch as well as potentially impacting on ventricular performance by inducing changes in afterload. To facilitate the diagnosis of abnormal or high-risk cases, an average reference model of the aortic arch is necessary. Such a representative mean shape of the patient population (the “atlas”) helps to quantify shape variations in 3D and allows statistical correlation of shape features to clinical outcomes. This retrospective study is based on a patient population of n=35 children with HLHS who underwent surgical arch reconstruction and subsequent cardiovascular magnetic resonance (CMR) examination. Three-dimensional surface models of the surgically enlarged aortic arches, excluding the brachiocephalic arteries, were reconstructed from the CMR whole-heart sequence serving as input data for the atlas calculation, Fig. 1. The atlas was calculated using a diffeomorphic surface registration algorithm based on a forward model, in which the surfaces are represented non-parametrically by currents [1]. Principal Component Analysis (PCA) was applied to the resulting shape vectors to determine dominant shape features in our population. The deviation of each patient’s aorta from the atlas was calculated as a dimensionless number in the space of currents. Fig. 1 shows the input surface meshes and the resulting atlas. Ten principal deformation modes from PCA accounted for 90% of the shape variation. Remarkably, 4 out of 5 patients with poor clinical outcome (defined as either death or cardiac transplantation) showed deviations from the atlas above the median value, Fig. 1. We present the first 3D shape atlas of the reconstructed aortic arch in patients with palliated HLHS. Our data suggest that clinical outcomes can be associated with shape features. The calculated PCA modes allow investigating this correlation more in detail. Results from such analyses can improve risk assessment and consequently affect surgical decision making in the completion of aortic arch reconstructions.