As a capstone to several Mid-America Earthquake Center (MAE Center) projects, a full-scale two story unreinforced masonry (URM) building was tested following the application of several retrofit techniques, which included the use of fiber reinforced polymer (FRP) overlays, near surface mounted (NSM) rods, vertical unbonded post-tensioning, and joist anchors. The test structure was composed of four URM walls, flexible timber diaphragms and interior stud walls, and was designed and built following construction practices consistent with those used in Mid-America prior to 1950. Initial testing subjected both the roof diaphragm and in-plane walls to slowly applied lateral load reversals in an unreinforced sate. Following this series of tests, each in-plane wall was retrofit and retested. Experimental results indicated that global issues such as flange participation and the effects of overturning moment substantially impacted the response of primary components both before and after retrofit. FRP retrofit techniques resulted in strength increases up to 32% and displayed a pseudo-ductile response caused by progressive debonding. For cases where such retrofits forced sliding failures, large increases in energy dissipation resulted. The use of vertical unbonded post-tensioning resulted in strength increases between 40%-60%; however, piers displayed a tendency to switch from a ductile rocking/sliding mode to a more brittle diagonal tension mode. In addition, results highlighted the need for retrofit schemes to employ both horizontal and vertical reinforcement to prevent progressive crack opening that can decrease wall displacement capacity. Based on the experimental results, the model implied by the and quot Prestandard for the Rehabilitation of Existing Structures and quot, FEMA 356, for the analysis of in-plane URM walls was modified and extended to (1) include the effect of FRP pier retrofits and (2) consider the global effects of URM structures. The resulting model displayed reasonable estimates of measured response both before and after retrofit, with an average error of 14%. In addition, the proposed model displayed improvements over the current model from 14% to 66%. Based on the results of sensitivity analyses this improved accuracy was primarily attributed to the consideration of global effects.