Thermal cracking in pile caps caused by concrete hydration heat will affect the safety and durability of long-span cable-stayed bridges. Therefore, effective prediction and control of concrete bridges hydration heat has been a challenging problem. In this study, the temperature of hydration heat in mass concrete pile caps belonging to a long-span cable-stayed bridge in China were monitored. Then, we adopt support vector machine regression (SVR) to establish the correlation between influencing variables and the temperature of hydration heat. The monitoring data are used to train to realize the short-term prediction of concrete temperature. The predicted results show that the SVR has a high accuracy, and the deviation between the prediction results and the measured values is quite small. The prediction performance of SVR for temperature of hydration heat of mass concrete is obviously better than that of BP neural network. The SVR prediction model can predict the temperature of 2-3 days with high accuracy. Based on the prediction results, temperature control method can be taken in advance to reduce the possibility of thermal cracks, which is of great significance for the safety and durability of actual engineering construction.INDEX TERMS Mass concrete, heat of hydration, support vector machine, regression model, temperature prediction.