In this work, a disturbance observer and state observer based terminal sliding mode (TSM) controller with acceleration data fusion is proposed for the active control of a seat suspension. In practical applications, the driver's body and the friction forces are difficult to be accurately described with a mathematical model; for this reason, the proposed controller is designed based on a model simplified from a 6-degree-of-freedom (6-DOF) seat-driver model with nonlinear friction. The disturbance observer and state observer are designed together with Linear Matrix Inequality (LMI) method. For improving the observer's performance, a complementary filter is applied to fuse the estimation of the seat suspension velocity from the acceleration measurement and the state observer. The proposed controller is validated using simulations with various bump excitations applied, and the conventional state feedback TSM controller is implemented for comparison. The proposed controller is also implemented in a practical active seat suspension prototype, and a well-tuned commercial heavy duty vehicle seat suspension is applied for comparison. The power spectral density (PSD) value and ISO 2631¿1 standard are used to evaluate the active seat suspension system's performance under random vibration. Both the simulation and the experimental results indicate that with the proposed controller, the vibration magnitude caused by a rough road can be greatly reduced, and the driver ride comfort is greatly improved.