Automated simulation of arbitrary, non-planar, 3-D crack growth in real-life engineered structures requires two key components: crack representation and crack growth mechanics. A model environment for representing the evolving 3-D crack geometry and for testing various crack growth mechanics is presented. Reference is made to a speci"c implementation of the model, called FRANC3D. Computational geometry and topology are used to represent the evolution of crack growth in a structure. Current 3-D crack growth mechanics are insu$cient; however, the model allows for the implementation of new mechanics. A speci"c numerical analysis program is not an intrinsic part of the model, i.e. "nite and boundary elements are both supported. For demonstration purposes, a 3-D hypersingular boundary element code is used for two example simulations. The simulations support the conclusion that automatic propagation of a 3-D crack in a real-life structure is feasible. Automated simulation lessens the tedious and time-consuming operations that are usually associated with crack growth analyses. Speci"cally, modi"cations to the geometry of the structure due to crack growth, remeshing of the modi"ed portion of the structure after crack growth and reapplication of boundary conditions proceeds without user intervention.