The purpose of this research was to use the finite element method as a comparative tool to assess the initial stability at the fusion site of ankle arthrodesis constructs internally fixed with screws. Digital image-based finite element models of ankle arthrodesis constructs were built and subjected to the load cases most likely to affect the ankle in the postoperative period. Mesh refinement and sensitivity analyses were performed upon a reference model. Two joint surface preparation techniques (preserved and flat cut joint contours) were compared in ankle constructs fixed with two and three screws in different configurations. Normal and poor bone quality were simulated. The initial stability was assessed by measuring the relative micromotions between the tibia and the talus at the fusion site, as well as examining the bone strain distributions. Preserved joint contours predicted better initial stability at the fusion site and a more uniform distribution of loads in the bones than resected joint surfaces. With two-screw fixation, regardless of the surface preparation technique, inserting the screws at 30 degrees relative to the long axis of the tibia and crossing them above the fusion site predicted the best performance. Increasing the insertion angle and lowering the crossing level caused an overall decrease in the stability at the fusion site. Three-screw fixation predicted better initial stability, a more uniform distribution of loads in the bones and a lower risk of bone failure than two-screw fixation. Adding a third screw anteriority produced, overall, a better performance than adding the screw posteriorily. As compared to the normal quality bone, large decreases in stability at the fusion site were predicted when different levels of poor bone quality were simulated. Even the most stable two-screw configuration predicted low levels of stability and regions at risk of bone failure, suggesting the need for a third screw.