Despite the extensive literature on reinforced concrete (RC) members retrofitted with fiberreinforced polymer (FRP) composites, few studies have employed a reliability-based approach to evaluate the seismic performance of RC buildings in terms of their collapse capacity and ductility. In this study, the performance of a poorly-confined RC building structure is investigated for different FRP retrofitting schemes using different configurations and combinations of wrapping and flange-bonded FRPs, as two well-established techniques. A nonlinear pushover analysis is then implemented with a computational reliability analysis based on Latin Hypercube Sampling (LHS) to determine the collapse capacity and ductility of the case-study structure. The variations in material properties and applied loads are examined using a rational probabilistic procedure. The results demonstrate the effectiveness of the reliability approach as it is capable of providing reliable and accurate comparisons between the retrofitting schemes implemented. In addition, the failure modes of the original and retrofitted frames are scrutinized for a more detailed study. It was found that the failure mode of the casestudy building is remarkably dependent on the variations of both the input parameters and the adopted retrofitting scheme.