A lot of attempts and investigations have been done in recent years to improve the structural seismic design. The main approach governs on these attempts has been the control of structural performance during earthquake. During the large earthquakes, most of the structures reveal nonlinear behaviour. So to estimate the seismic performance of a structure, it is necessary to develop new methods which are based on nonlinear analysis. It seems that the nonlinear dynamic analysis is the most appropriate method, though it suffers from complexity. Therefore, the modern concept of "performance based seismic engineering” was developed, in which; we mostly use simple nonlinear static analysis to determine the nonlinear response of the structures. However, this method has also some weak points. One of these weaknesses is that only the first mode controls the whole response. And the other is that this mode remains unchanged after the structure under goes to inelastic range. Both of these assumptions are approximate, especially in tall buildings. It seems that the invariant lateral loads used in this method cannot consider these effects. In this paper, the accuracy of the invariant lateral load pattern in FEMA and modal pattern used in nonlinear static method and MPA method are considered in comparison with nonlinear response history analysis. This paper focused on drift and displacement as results of structure. These considerations were applied to three Steel moment EBF frames: 3, 9 and 15 stories and their detailed analysis were reported.