Relativistic electrons were observed up to tens of MeV energies in the 1970s and 80s (Datlowe 1971, Moses et al. 1989), but in the past 30 years, there have been no measurements at those high energies to compare with the much more sophisticated observations of ions and electromagnetic emission. In contrast to keV - tens of keV electrons, relativistic electrons generally exhibit a diffusive flux-time profile, indicating that substantial scattering has occurred in their propagation to ~1 AU. Electrons are as fundamental as ions to the understanding of the energy release process. Their spectral shapes provide a nearly perfect diagnostic of the SEP event type (Moses et al. 1989) - gradual events show single power laws in momentum, while impulsive events all show a spectral hardening starting at around 5 MeV/c. Although the hardening of the spectrum is universally observed, the spectral indices vary from event to event. Another surprising result (from combined Helios/ISEE-3 electron spectra) is that the spectral shapes are apparently invariant as a function of the azimuthal distance to the flare if the fluxes are adjusted for different radial distances. Little is known as to the origin of these electrons or the reason for the hardening. The double power law spectra have been discussed in terms of a superposition of two electron populations, one accelerated in flaring loops by a stochastic mechanism, and the other by a shock in the high corona (Dröge 1996a). How shocks can accelerate electrons to relativistic energies (never observed for shocks near 1 AU) near the Sun is still an unsolved mystery: can coronal shocks accelerate electrons to relativistic energies starting from a quasi-thermal population, or is a more energetic seed population necessary?