Active engine mounts provide an effective solution to further improve the acoustic and vibrational comfort of vehicles with emission-reducing engine technologies, e.g., cylinder-on-demand. To control active engine mounts either adaptive or non-adaptive feedforward control is commonly employed. Since both approaches previously have been treated separately, this thesis proposes methods to connect them in terms of adaptive filters with self-trained grid-based look-up tables. By incorporating the two control strategies, their inherent advantages, i.e., the adaptivity of adaptive filtering and the direct impact as well as the tracking behavior of map-based feedforward control, are combined. In addition, the joint consideration of the two feedforward control structures provides a viable approach for data generation of map-based feedforward algorithms. The successful application of adaptive and map-based feedforward control mostly depends on the controlled plant and its respective variations during vehicle operation. Large variations may degrade the performance or even destabilize adaptive feedforward algorithms. In this case stabilizing countermeasures, e.g., online system identification, are necessary. On the other hand, map-based feedforward control becomes ineffective in the presence of transfer path variations. Therefore, in the second part of this thesis, the active and passive characteristics of an active engine mount as well as their variations in a vehicle environment are investigated. The analytical and experimental results emphasize the importance of the active engine mount actuator's resonance frequency for the use of adaptive and non-adaptive feedforward control algorithms. Map-based feedforward control can only be effectively employed in frequency regions well below the actuator's resonance frequency, where only small transfer path variations occur. However, adaptive feedforward control can be operated within the resonance region due to a proposed novel online subband identification scheme. Finally, in-vehicle tests show a superior vibration cancellation of the adaptive feedforward control algorithm compared to map-based feedforward control. Nevertheless, the latter still provides a significant, subjective perceptible vibration reduction, despite the variability of the analyzed six vehicles.