By the recent researches of the shaking table tests of wooden post-and-beam houses, the effect to the seismic performance by exterior walls was not negligible. However, it is difficult to estimate their effect to the seismic performance of structural system precisely, when the exterior panels straddle multiple stories or shear walls. Moreover, if the exterior walls are made by wet construction (e.g. mortar plastering), the analysis of seismic performance of whole structure is more complex, because the composite effect of plastered wall according to plastering area cannot be ignored. By the analysis of the load-displacement relationships of the shaking table test conducted at E-defense (NIED, Miki-city, Japan) on 2005, it was found that the story shear of the experimental result of earthquake input was much higher than the estimation of the story shear by the summation of the results of the static shear tests of the wall element used in shaking table specimen. This margin was estimated to be due to some differences between static tests and dynamic shaking table tests (i.e. dynamic loading effect, twist behavior). Some analytical approaches for this shaking table tests indicated that it was because of the multi-story behavior of mortar wall or the underestimation of shear performance of structural elements. It was important to understand the effect to the seismic performance by exterior walls for the precise estimation of the seismic performance of real wooden houses.<br> In this paper, the new numerical analysis model for exterior walls like mortar walls in Japanese post-and-beam wooden houses was suggested and the reproductive numerical analyses of the shaking table tests conducted at E-defense were carried to estimate the effect to the seismic performance by exterior walls. The two shaking table test specimens that had the mortar plastered exterior walls and mud plastered interior walls were used as target specimens. In the analytical model of them, shear performance of the shear walls and tensile performance of the joints between frames were decided by the result of element tests and the test of literatures. The two modeling methods for the exterior mortar walls were used. One is the brace substitution (BR-model) and the other is consists of the panel elements and linked springs (PN-model). The panel elements were rigid and connected to the wooden frames by two-direction shear spring. The neighboring panel elements were linked each other by the link springs. By this modeling method, the composite effect of plastered wall according to plastering area was realized.<br> By the numerical analysis of the lateral loading for shear walls and the shaking table tests conducted at E-defense, followings was found.<br> 1) The numerical analysis results of PN-model for the lateral loading of shear walls were agreed with the theoretical results by the calculating method that estimate the rigidity and the yield shear force of shear walls.<br> 2) By comparison of BR-model with PN-model, the multi-story effects and size effect of the mortar walls increase the story shear of the first-story by 20 to 60% in push-over analysis and by 15 to 55% in time-history response analysis.<br> 3) By the numerical analysis of shaking table tests, it was found that the multi-story effects and size effect of the mortar walls were realized by our modelling method.<br> 4) The upper deformation of column end joints of the numerical models was smaller than the experimental results. The improvement of skeleton curves of springs for analytical models is needed.