Following myocardial infarction, damaged muscle is gradually replaced by collagenous scar tissue. The structural and mechanical properties of the scar are critical determinants of heart function, as well as the risk of serious post-infarction complications such as infarct rupture, infarct expansion, and progression to dilated heart failure. A number of therapeutic approaches currently under development aim to alter infarct mechanics in order to reduce complications, such as implantation of mechanical restraint devices, polymer injection, and peri-infarct pacing. Because mechanical stimuli regulate scar remodeling, the long-term consequences of therapies that alter infarct mechanics must be carefully considered. Computational models have the potential to greatly improve our ability to understand and predict how such therapies alter heart structure, mechanics, and function over time. Toward this end, we developed a straightforward method for coupling an agent-based model of scar formation to a finite-element model of tissue mechanics, creating a multi-scale model that captures the dynamic interplay between mechanical loading, scar deformation, and scar material properties. The agent-based component of the coupled model predicts how fibroblasts integrate local chemical, structural, and mechanical cues as they deposit and remodel collagen, while the finite-element component predicts local mechanics at any time point given the current collagen fiber structure and applied loads. We used the coupled model to explore the balance between increasing stiffness due to collagen deposition and increasing wall stress due to infarct thinning and left ventricular dilation during the normal time course of healing in myocardial infarcts, as well as the negative feedback between strain anisotropy and the structural anisotropy it promotes in healing scar. The coupled model reproduced the observed evolution of both collagen fiber structure and regional deformation following coronary ligation in the rat, and suggests that fibroblast alignment in the direction of greatest stretch provides negative feedback on the level of anisotropy in a scar forming under load. In the future, this coupled model may prove useful in computational design and screening of novel therapies to influence scar formation in mechanically loaded tissues.