Atmospheric temperature and directed solar radiation have a significant effect on the temperature field of high-speed railway (HSR) concrete bridge and ballastless track structure. However, temperature actions are random process of which distribution laws are difficult to explore, and existing statistical methods for structure temperature analysis are still not precise. So far, there are few researches for annual temperature spectra and design codes for bridge–track system. Based on the one-year observation data, this paper investigated the temperature actions for Chinese HSR bridge–track structure. By utilizing reliability high moment theory, a statistical method which could built virtual distribution was put forward. Based on the renewed study, the effects of waterproof for deck were taken into consideration, a temperature action model was proposed which is suitable for both bridge and track structure. In addition, for track structure, the previous temperature load models were modified. Apart from that, by proposing the concepts of temperature uniform and fluctuant spectra, the research evaluated service performance of structure. Finally, the distribution regularities of uniform temperature spectra were fitted by Fourier series, and the relationship between structural and atmospheric uniform temperature was established (formula (25)). As a result, according to 50 years recorded atmospheric temperature data, the prediction model of the structure extreme temperature was suggested, and by taking the recurrence interval of 100, 150 and 300 years, the extreme temperatures of the system are 52.23, 54.34 and 57.77 °C.