Small-scale physical model testing can be applied to calibrate a computational building energy model based on the acquired reliable thermal performance data. The calibration approach using a small-scale physical model is a rigorous and cost effective means in order to do an accurate full-scale building energy simulation. It is meaningful to reveal the correlation of the thermal performance between a full-scale building and its scale model, so that the data from a small-scale model testing can be scalable to predict the performance of a full-scale building. The purpose of the research is to investigate the correlations reflecting the comparative thermal performance of a full-scale building against its scale model with identical constructions and geometrical similarity. More than ninety thousand scenarios about the typical target building zones and their corresponding scale models were generated via the parametric energy modeling and simulation approach. Then, the typical thermal performance indicator of each building zone, i.e. the zone mean air temperature (ZMAT), was obtained. The result shows that the zone mean air temperature difference between the target building and its scale model is a time dependent variable with the diurnal cycle. Compared with a fixed scale model, the bigger building causes the larger variation of zone mean air temperature difference. Furthermore, it also indicates that the zone mean air temperature difference can be approximated by a logarithmic regression function as ALn(SF/d) + B, where A and B are empirical coefficients associated with climate, building form, material and construction, as well as window-towall ratio. In engineering practice, the outcomes of the research are applicable for thermal performance estimation and energy model calibration of full-scale buildings, especially for a planned building with the available material and construction samples, but without the determined or reliable thermal performance data for the building.