The curtain wall is being widely adopted in high-rise buildings because of its many advantages. Many researchers have tried to predict the surface temperature of windows or curtain walls precisely. But most of the research has focused on small-scale window rather than large-scale curtain walls. Thus, a study on the correlation between a computer simulation and a mock-up test using a large-scale thermal chamber for the curtain walls is needed. There are three main objectives in this study: first, to understand the reasons for the differences between the results of the thermal mock-up tests and those of the simulations: second, to understand the differences in the simulation methods: and third, to reduce those differences using the method proposed in this study. For 12 curtain walls, the mean temperature difference between computer simulations and the mock-up tests was about 2.6 degrees C. The simulation method (NFRC) considering convection and radiation showed closer result to the mock-up test than conventional simulation method. This study proposed a method adjusting the simulation result to the test result by changing the convective film coefficients of the frame and glazing to find the optimal convective film coefficients for the chamber. To verify proposed the average convective film coefficients, additional thermal mock-up tests and simulations were performed. The simulation using the average convective film coefficients showed better agreement with the mock-up result. (C) 2007 Elsevier B.V. All rights reserved.