Resonant electromagnetic actuators have been broadly used as vibration motors for mobile devices given their ability of generating relatively fast, strong, and controllable vibration force at a given resonant frequency. Mechanism of the actuators that is based on mechanical resonance, however, limits their use to a situation where their resonant frequencies are known and unshifted. In reality, there are many factors that alter the resonant frequency: for example, manufacturing tolerances, worn mechanical components such as a spring, nonlinearity in association with different input voltage levels. Here, we describe a sensorless resonance tracking method that actuates the motor and automatically detects its unknown damped natural frequency through the estimation of back electromotive force (EMF) and inner mass movements. We demonstrate the tracking performance of the proposed method through a series of experiments. This approach has the potential to control residual vibrations and then improve vibrotactile feedback, which can potentially be used for human-computer interaction, cognitive and affective neuroscience research.