Objective: To create a comprehensive finite element model capable of analyzing the biomechanics of canine retraction. Methods: A half maxilla virtual model with an extracted first premolar was created from human computed tomography data. Accurate brackets and an 0.018 archwire were placed to model canine retraction under 0.5N and 1.0N of retraction force. A two-tooth substructural model was isolated to examine the importance of surrounding geometry. Additionally, mesh size and periodontal ligament (PDL) elastic modulus were varied to examine the effect on predictions. Comparisons were made to previously published clinical data. Results: The substructural model decreased computational load, but altered maximum stress up to 16.4%. Coarse mesh sizing affected displacement results up to 22% and maximum stress up to 47%. No PDL stiffness was able to accurately represent the clinical data. Conclusions: Modeling canine retraction was partially achieved, highlighting the importance of mesh sizing and the need to incorporate remodeling.