The improvement of exhaust emission during engine warm-ups is vital in engine emission control as engine emission limits are constantly lowered. An effective solution to this problem is to install a rapid-warming catalyst. On the other hand, precaution also has to be taken to avoid overheating of the catalyst. These require detailed information on heat transfer and accurate gas temperature variations at different locations in the exhaust system. In this study, experiment was conducted to investigate how mean gas and pipe wall temperatures vary during warm-ups throughout the exhaust systems, as well as the time constants of the transient processes. In addition, a program was developed for the simulation of exhaust gas and pipe wall temperatures during warm-ups. The temperatures were time-averaged in every engine cycle but variable from cycle to cycle. Based on the experimental data, a number of existing models for unsteady heat transfer between the gas and pipe wall in manifold systems were examined and compared. Such an investigation suggested that the Reynolds analogy model, when modified to include the radiation heat transfer between the pipe wall and its surroundings as well as the wall and exhaust gas, could give the best temperature predictions. The comparison of the calculated and measured time-dependent gas and wall temperatures showed that the predictions were basically correct although certain discrepancies still exist, especially at the upstream ends of the exhaust pipes. From this study, it is concluded that the simulation package is a useful tool for the design of rapid-warming exhaust exhaust systems. Among models for heat transfer, the Reynolds analogy supplemented by the radiation heat transfer models is most capable of producing the best temperature predictions during engine warm-ups. This is especially true if temperature of the pipe wall becomes high enough. It is proposed that the longitudinal heat conduction in exhaust pipes near the ports, which was neglected in this study, should be considered if further improvement in temperature predictions is to be achieved.