This dataset is part of the publication "A bi-atrial statistical shape model for large-scale in silico studies of human atria: model development and application to ECG simulations" (C. Nagel et al., doi:10.1016/j.media.2021.102210). It includes a bi-atrial statistical shape model built based on 47 MR and CT images (Left atrium segmentation challenge (Tobon-Gomez, 2015), Left atrium fibrosis and scar segmentation challenge (Karim, 2013), Left atrial wall thickness challenge (Karim, 2018)). ScalismoLab (https://scalismo.org) was used for parts of the model generation. Further Details are explained in the paper. The SSM is available as an h5 file including information about the mean shape's vertex locations and their triangulation as well as the eigenvectors and -values. 100 random instances derived from the model are also available as vtk files. They were augmented in a post-processing step with a homogeneous wall thickness, anatomical labels, intra-atrial bridges and material tags so that they are ready to use for electrophysiological simulations of atrial signals.