Bearings are important components in rotating machines. An initial small damage in the bearing may cause a fast degradation, which may lead to the machine breakdown. The health condition of bearings can be monitored using proven vibro-acoustic methods effective for detecting bearing faults. However, the existing bearing health indicators do not provide a reliable estimation of the fault characteristics, such as fault size and fault location. As a result, the ability to assess the severity of the bearing damage and to make maintenance decisions is limited. The presented study is a part of an ongoing research on bearing prognostics, aimed to improve the understanding of the effects of fault size on the bearing dynamics. The research methodology combines dynamic modeling of the faulty bearing with experimental validation and confirmation of model simulations. In the presented study, small faults (starting from 0.3 mm), simulating incipient damage are generated at increasing sizes by an electrical discharge machine. The recorded vibration data is then analyzed and compared to the vibration signatures predicted by the model. The experimental and the simulation results add new insights on the manifestation of the size of the fault and possible indicators of the damage severity.