Non-linear time history earthquake response of reinforced concrete multi-story moment resisting frame structures are calculated with a model of beam-column joints considering their restoring force characteristics which represents the interaction of moment, axial force and shear (Figure 1). The influence of the failure mechanism and strength degradation of beam-column joint are investigated by the comparison of the simulation of twenty-seven model frames (Table 1) satisfying the current seismic provisions in Japan, the model parameters for which are combination of column-to-beam strength ratio of 1.2, 1.5, and 2.0 and joint hoop reinforcement ratio of around 0.2% to 0.6% which are neglected as design parameters affecting the seismic safety in design practice in Japan. Statically cyclic response analyses of the model frames subjected to lateral load distribution pattern of Ai by displacement control at the top floor are calculated to classify the type of hysteresis loop and strength degradation (Figure 3). Then dynamic analyses of the model frames are carried out subjected to five artificial base ground motions conformed to maximum considered intensity design (MCD) spectrum and five observed strong motions (Figure 4). It is found that there is significant difference in the story drift response of the structures with different beam-column joint attributes. Particularly, in some cases, the story drift ratio increased significantly due to concentration of lateral deflection to several stories and large residual drift remains to one direction (Figure 5). It is confirmed that the differences are due to softening of stiffness and strength degradation of the beam-column joints. It is caused by the mechanism identified as joint hinging (ref. 1) which features the yielding of longitudinal reinforcement within the beam-column joint and repetition of diagonal crack open and close. The state-of-the-practice analytical model for frames does not consider the joint hinging so it significantly underestimates the story drift responses by far than seismic design criteria (Figure 8). The response of frames with beam-column joint failing in joint hinging is very sensitive to base ground motions which were synthesized with same target response spectrum. So the prediction of the response is difficult by non-linear static procedure nor non-linear time history procedure the model of which consider plastic deformation only by non-linear rotational springs at beam-ends considering hysteresis model with slip behavior. The results of incremental dynamic analysis (IDA) shown in Figure 9 reveals that the amplification less than 50% of MCD yields little difference in the response of the models whilst the amplification of 100% of MCD causes near collapse to four-story model and concentration of story drift exceeding 6% to the models failing in joint hinging mechanism, which is far beyond seismic design criteria for MCD. The current Japanese seismic design practice for beam-column joint obviously lacks the necessary reinforcing detailing provision such as minimum column-to-beam strength ratio and design of required joint hoop ratio which is much behind the seismic provisions found in current NZ3101 and ACI 318 building codes.