The physical model of shear transfer mechanism in reinforced concrete beam-column joints in which the New Zealand Concrete Structures Standard NZS 3101: 1995 is based provides a good insight into the seismic behaviour of joints. However there are still some issues observed in the laboratory work that can not be fully explained with such model. This research project is aimed at improving the understanding of the seismic behaviour of joints. The research work seeks the endorsement of the design recommendations for interior joints given by the Concrete Structures Standard. The lower bound theorem of plasticity was applied to find the internal force trajectories within the joint panel. The diagonal compressive stress field of the joint was modeled with variable angle struts-and-ties. Relative importance of parameters influencing the shear strength of the joint panel was identified through the series of parametric analysis. The database consisting of 60 tests were processed to be used in conjunction with the analytical work. It was found that a clear trend exists between the ductility of the frame subassemblies and the joint shear stress ratios equivalent to a reference joint. This relationship was used to derive the design recommendations for the requirements of horizontal joint shear reinforcement of joints of ductile frames and limited ductility frames. An experimental programme was conducted to validate the analytical results with particular emphasis given to parameters that were found to be in disagreement between the analysis and the current design recommendations. Eight cruciform subassemblies were tested under simulated earthquake loading. Precast concrete was incorporated in the fabrication of the test units to simulate the design practice. There are five units in which beam bars are lumped at the top and bottom beam chords, while three units incorporate distributed longitudinal beam reinforcement. Grade 500 reinforcing bars were used as beam and column longitudinal reinforcement in all units. Test results showed good agreement with the analytical model within reasonable accuracy. Some of the important findings are summarized below. First, column compressive loads are not always beneficial to the joint strength. When the column axial load level exceeds 0.3fcAg, it becomes detrimental to the joint. Second, according to the results obtained in this study, the design recommendations given by NZS 31 O 1: 1995 are conservative in general and could be relaxed, except for some rare cases. Third, the horizontal joint reinforcement is strongly influenced by the ratio Vjh / f c rather than by the bond force of the longitudinal beam bars. Forth, the requirement of horizontal joint reinforcement given by NZS 3101: 1995 for joints in which the amount of top and bottom beam bars is unequal was found to be unduly stringent. Fifth, the shear strength of the joints in which beam bars are distributed along the web is very similar to that of the conventionally reinforced joints. Therefore, no relaxation of amount of horizontal joint reinforcement can be expected when using this design alternative. Test results showed that the theoretical model established in this study is able to predict the joint strength in correlation with the ductility. Joint design procedures based on the traditional forced based and displacement based design are discussed in this work. The effect of using high-grade reinforcement on the bond strength within the joint is also studied. Test results and theoretical predictions conclusively showed that the yield drift of the frame subassembly becomes large when Grade 500 longitudinal reinforcement is incorporated. As a result, full ductility can seldom be achieved before reaching the interstorey drift limitation of 2.5% given for the ultimate limit state by the loadings code, NZS 4203: 1992. Drift limitations are expected to control the design of reinforced concrete moment resisting frames when Grade 500 reinforcement is used as longitudinal bars in columns and beams. It is suggested that, except for low-rise structures in which the drift limit can be easily met, moment resisting frames designed using Grade 500 bars be designed only for limited ductility response.