Recently, research and development on structural health monitoring (SHM) technology has been advanced for the purpose of proper maintenance and rapid diagnosis of disaster. In order to disseminate this technology, it is necessary to reduce the price of the whole system including installation and maintenance costs. One solution is to promote wirelessization of a system that makes it unnecessary to lay cables to each part of the building. In this case, there is a problem of how to secure the simultaneity of the data recorded with multiple vibrometers. In this paper, a simple method is proposed for synchronous correction of input / output waveforms recorded asynchronously in structural health monitoring. The main feature of this method is that by identifying the transfer function of the target building from the acceleration record and converting either the input or the output waveform by the transfer function, the accuracy of correlation evaluation (detection of synchronous deviation) between waveforms. In this study, applicability of the proposed method was verified by using multiple numerical examples in which pseudo synchronization deviation and noise were added, for two-story housing. Two estimation models of the transfer function were proposed, one with using 2 mass system model with the same degree of freedom as the target building and the other using 1 mass system model easy to handle, and compared the estimation accuracy. The noise of the input / output waveform was taken into account by adding independently generated band limited white noise to each waveform. The noise level (the ratio of RMS value of noise waveform to that of original waveform) was varied from 0 to 100%. The main results of this study are as follows. (1) Transfer functions of 1 and 2 mass system model were expressed explicitly with the variables of natural frequency and damping constant. (2) In the estimation of the transfer function, it was confirmed that the transfer function without synchronization deviation can be accurately estimated by identifying the parameter that minimizes the error with the above model with the amplitude ratio of the input / output waveform as the target. This is based on the fact that when there is a synchronization deviation between the input and output waveforms, the phase difference is greatly disturbed, while the influence on the amplitude ratio is small. (3) By estimating the transfer function using 1 mass system model, it was confirmed that it is possible to estimate the synchronization deviation stably and accurately even if there is noise in the recorded waveform. In addition, it was confirmed that synchronization deviation can be estimated with practical enough accuracy even by using seismic observation records of two-story real housing.