Dynamic thermal monitoring at of IGBT modules is a key technology to enhancing the reliability of high-power converters. However, most existing thermal model-based methods suffer from temperature estimation errors due to variations in model parameters and inaccuracies in loss calculations. To address this issue, this paper proposes an adaptive Kalman filter-based 3-D thermal monitoring method for IGBT modules. Firstly, a 3-D reduced-order thermal model for the IGBT module is established, enabling real-time estimation of temperatures. Secondly, a feedback loop is introduced using an adaptive Kalman filter to correct the thermal model. Simulation results demonstrate that, when confronted with deviations in thermal model parameters and loss calculations, the proposed adaptive Kalman filter-based thermal monitoring method can improve temperature estimation accuracy compared to open-loop thermal models.