In this paper, a quasi-continuous sliding mode control is applied for the tracking control of a hydrostatic transmission. The nonlinear dynamic model of the considered system consists of four first-order differential equations characterised by both parametric uncertainty due to uncertain physical parameters and unknown disturbances, the leakage volume flow as well as a resulting load torque. A nonlinear reduced-order state and disturbance observer is employed to estimate the unmeasurable system states and disturbances online. The proposed control structure is validated both by simulations and experiments. A comparison with flatness-based control shows that a superior tracking accuracy can be achieved for the normalised motor bent axis angle and the motor angular velocity as controlled variables.