Steering motors are involved in the vehicle active-steering or motor-aided steering systems such as active front wheel steering, steering-by-wire, and power steering systems, and steering motor faults always impose a great threat to vehicle stability and safety, demanding dedicated fault estimation systems. There are several challenges existing in the fault estimation for steering systems: time-varying vehicle longitudinal speed, inevitable modeling errors, and time-varying or high-order faults. In order to cope with all these challenges, a robust fault estimation approach is proposed by a means of gain-scheduling H∞ strategy. First, the vehicle model is established and incorporated with the steering system fault model. Then by taking into consideration of parameter-varying and modeling errors, a robust estimator is presented that ensures H∞ performance from modeling error to fault estimation. Experimental validations are shown to verify the effectiveness of the proposed estimator.