Autonomous driving of mining trucks is crucial to the construction of intelligent mines. In open-pit mines, complex terrain changing brings great challenges to the control performance of autonomous trucks. To solve this problem, a novel terrain-adaptive longitudinal control (TA-LC) algorithm is proposed by combining the adaptive feed-forward PID and the mine terrain. The TA-LC requires fewer calibration parameters, which only need one set of parameters to keep stability under various terrains. Compared with the traditional PID controller, the steady-state error of the vehicle speed is reduced. Simulation experiments are carried out to verify the effectiveness and stability of the TA-LC. The TA-LC is also applied on the autonomous trucks in JiDong Cement Mine.