Perennial ice deposits in caves are an underexplored component of the cryosphere preserving a largely untapped archive of long-term changes in landscape and climate whose existence is threatened by climate change. This study demonstrates how terrestrial laser scanning (TLS) can be used to fully and accurately (registration accuracy < 1 cm standard deviation of point differences) assess the geometry of an ice-bearing cave in the Eastern Alps (Tyrol, Austria). Three TLS campaigns and 255 scan positions were used to acquire point clouds with a high sampling density (2 cm average point spacing) in order to minimise shading effects and to assure a precise and highly resolved 3D documentation of the cave. A semi-automated registration and point cloud-processing approach adapted to the site-specific demands ensured a complete and error-minimised assessment of the cave's geometry serving as a solid basis for future quantifications of snow and ice content dynamics. Dominant cave surface structures were investigated by performing a multiscale principal component analysis (PCA) to identify a detailed and computationally efficient basis for future airflow modelling tasks.