We address the problem of rotational velocity and acceleration estimation from incremental encoders in the presence of sensor imperfections, with a particular orientation towards automotive applications such as traction and brake control. In this area, measurements of the wheel speed and acceleration are often affected by large disturbances with a period of one revolution that arise from sensor imperfections and degrade the performance of most control designs. We present an algorithm to identify and remove such periodic perturbations online, without the need of error compensation look-up tables, and without assuming constant velocity. Experimental tests prove that the method is able to greatly reduce the impact of perturbations without introducing an important phase lag, as opposed to the results obtained using a notch-filter. Index Terms-Velocity and acceleration estimation, in-cremental encoders, periodic measurement noise, time-stamping algorithm, parameter estimation, antilock braking system.