Infrared thermography is an effective method for assessing the on-site thermal performance and heat exchange of structures. On-site infrared thermography was generally performed under the assumption of quasi-steady sate, which refers to the averaged boundary conditions; however, a description of the actual on-site dynamic conditions is necessary to increase the reproducibility of the assessment. This study, through simulation models, describes the infrared thermal images of the surrounding wall depending on the thermal performance at the joint between the window and wall, through which the thermal performance of the joint is estimated to quantify insulation defects or thermal bridge flaws. As incremental research to the previous chamber-based preliminary research, the main objective was to apply the assessment method to an on-site experiment and verify the reproducibility of the results. In the case of the distinct differentiation of the temperature distribution of the wall according to the material properties of the joint, the reproducibility of the assessed results was verified as values of the material properties of the joint of the physical model that represent the minimum error were the same, regardless of the period of assessments. Furthermore, the difference in the results from the physical model, in which the derived physical property values were applied, and the infrared thermography measurements were also within 5% and 20% (coefficient of variance of the root mean square error and normalized root mean squared error).