Dielectric elastomers are highly promising as functional materials for the rapidly developing field of flexible actuator and generator technology. They offer a unique combination of low densities and large reversible deformations of up to more than 100% in area, and consequently have great potential for many new types of application. However, implementation has been impeded by the lack of specialized materials. The elastomers that have so far been investigated suffer from a number of disadvantages, including the need for very high activation voltages and limited service lifetimes. This thesis describes an investigation of the use of elastomeric composites as dielectric elastomers with the aim of optimizing and improving their performance in actuators. The influence of materials properties on actuation was first analyzed on the basis of a simple physical model and materials properties derived from standard test methods. The implications for the actuation performance of three conventional dielectric elastomers were then considered in detail. A preliminary conclusion was that the actuation performance could be improved if the permittivity of the elastomers were to be increased by modification with ceramic or conductive fillers. However, actuation performance was shown to depend not only on the permittivity, but also on the elastic modulus, the electrical breakdown strength, and strain hardening. Thus, although significant increases in permittivity were achieved by this approach, actuation performance was compromised by an increase in modulus in the case of the ceramic fillers, and a dramatic drop in electrical breakdown strength, in the case of the conducting fillers. A more promising approach was therefore suggested to be the use of an organic conducting filler encapsulated in an insulating matrix. It was demonstrated that it is indeed possible to increase the permittivity of a given elastomer while maintaining a high electrical breakdown strength. Different processing routes were investigated in order to control the dispersion of the filler and tailor performance. The optimum filler concentration, i.e. that providing the best compromise between permittivity and stiffness, was determined to be approximately 16 vol%, resulting in an improvement by a factor of 2 in actuation strain for a given applied voltage over that obtained with the unmodified matrix. Higher filler concentrations were also argued to have considerable potential for use in generators, given that the observed increased permittivity was also associated with high electrical breakdown strengths and increased strains at break. A threefold increase in converted energy per working cycle was predicted for a composite containing 25.5 vol% fillers based on a simplified model for a dielectric elastomer generator. Whilst these results are extremely encouraging, it is concluded that the composite approach has, in general, only limited potential as a means of obtaining further increases in actuation performance. The major difficulty remains that the use of a relatively rigid second phase to increase dielectric performance will inevitably also increase the elastic modulus beyond a certain filler concentration. As argued in the final part of the thesis, the way forward may therefore ultimately depend on the development of new types of synthetic elastomeric matrix materials that combine intrinsic improvements in electrical response with reduced moduli.