||External bonding of FRP plates to the tension substrate of RC beams has been accepted as an efficient and effective technique for flexural strengthening. In this thesis, different problems related to crack-induced debonding of the FRP plate in the flexural strengthened concrete beams have been investigated. FRP strengthened RC beam may fail by FRP debonding from the bottom of a major flexural crack in the span. This kind of failure is studied with the direct shear test in the present research work. Our experimental investigation focuses on the effect of concrete composition on the bond behavior between FRP and concrete. Based on the test results, the bond capacity of the specimen is found to be governed by the concrete surface tensile strength, aggregate size and aggregate content. Then, the neural network is employed to derive an empirical expression for the interfacial fracture energy in terms of concrete surface tensile strength and aggregate content. Using the empirical equation, simulated bond capacity is in good agreement with experimental results. Then, an analytical model has been developed to study the debonding behavior in the direct shear test. With proper interfacial parameters obtained from the fitting of test data, good agreements between the theoretical and experimental debonding behavior can be achieved. Debonding of the FRP plate from a major flexural/shear crack in the strengthened beam is also a common failure mode. For a flexural/shear crack, there are both horizontal and vertical displacements between the two sides of the crack. In the present study, both experimental and theoretical investigations have been carried out to study the effect of the vertical displacement on the debonding behavior in the FRP strengthened concrete beams. To predict the ultimate FRP strain for the mixed mode debonding failure, an analytical model has been developed based on a new shear slip relation with consideration of the vertical displacement. With the use of a single set of interfacial parameters, predicted debonding behavior in various members with different FRP thickness and different vertical displacements at the flexural/shear crack is found to be in good agreements with experimental results. The validity of the proposed modeling approach is hence verified. In the FRP strengthened RC beams, debonding of the FRP plate often occurs under the presence of multiple cracks along the span. In the present thesis, experimental and theoretical investigations are performed to study the effect of multiple secondary cracks on the debonding behavior and ultimate load capacity. A new analytical model for FRP debonding under multiple cracks has been developed. The effect of the multiple secondary cracks on the shear softening in the debonded zone is explicitly considered in the model. Using the new model, the simulated values of ultimate load when debonding occurs are in good agreement with measured values. In the FRP strengthened RC beams, concrete cover separation or plate end debonding can be avoided by applying tapers at the FRP plate end. In this situation, it is easier for FRP debonding to be induced by a major flexural crack close to the support. To study the effect of the taper end on the crack-induced debonding behavior, finite element analysis is first conducted to investigate the stress distributions along the concrete/adhesive interface and along the FRP plate. An analytical model for debonding of tapered FRP plate is then developed. The analytical and FEM results are in good agreement with one another. Using the analytical model, the effect of the taper configuration and FRP bond length on debonding behavior has been studied.