Soil temperature is a key driver of several physical, chemical, and biological processes. The Environmental Policy Integrated Climate (EPIC) is a comprehensive ecosystem model that simulates soil temperature dynamics using a cosine function approach driven by daily air temperature and average annual soil temperature at damping depth, which may erroneously predict lower soil temperatures in winter. A new cosine model and a pseudo-heat-transfer model were therefore developed and implemented for simulating soil temperature. The two methods were evaluated by comparing simulated daily soil temperatures with observed data at 24 study sites. Results showed that the two new methods had similar performance and the better statistical results obtained with these new methods demonstrated the ability to better predict the soil temperature for a wide range of pedoclimatic conditions, land management, and land uses. The main reason for the improved performance was due to a better prediction of soil temperature during the winter period.