This article focuses on algorithms for fast computation of the Euclidean distance between a query point and a subdivision surface. The analyzed algorithms include uniform tessellation approaches, an adaptive evalution technique, and an algorithm using B´ezier conversions. These methods are combined with a grid hashing structure for space partitioning to speed up their runtime. The results show that a pretessellated surface is sufficient for small models. Considering the runtime, accuracy and memory usage an adaptive onthe-fly evaluation of the surface turns out to be the best choise.