In the framework of RQMD we investigate antiproton observables in massive heavy ion collisions at AGS energies and compare to preliminary results of the E878 collaboration. We focus here on the considerable influence of the real part of an antinucleon-nucleus optical potential on thep momentum spectra.PACS numbers: 14.20 Dh, 25.70.-z Antibaryon production is a promising observable for collective effects in nucleus-nucleus collisions. On the other hand, N 's suffer strong final-state interactions. These interactions have two components which can be related to the N self-energy in matter: collisions and annihilation on baryons[1] (imaginary part, semiclassically given by 2ImV = σvρ) and a piece in the real part (ReV = t NN ρ in the impuls approximation). In the semiclassical limit the real part of the self-energy can be approximated by potential-type interaction[2] or a mean field. Here we will focus on the effect of the real part. The motivation is that the long-range force of baryons acting on ap is expected to be stronger than for protons since the Lorentz-scalar and the Lorentz-vector parts of a meson exchange potential now have the same sign. The influence of baryonic mean-fields on baryons and mesons is well established. Therefore there should also be some influence onp's. The substantial impact of mean-fields on particle spectra was studied earlier [3,4]. Following these ideas, we now investigate observables in nucleus-nucleus collisions, where BB-potentials come into play. For this purpose, we employed a simple modelinteraction, knowing that this choice is far from being unique. In principle, one has to calculate the medium-dependent mean-field and the crosssection selfconsistently to understand N behaviour in matter. We calculate the forces acting on a N in a baryonic medium only after freeze-out. However, by taking the free interaction -annihilation, elastic scattering, non-annihilating inelastic channels -for NN in the collision term during the dynamical evolution the real potential is included effectively. Our approach is similar in its spirit to the usual treatment of the Coulomb distortion of particle spectra which is also restricted to final-state interactions [5]. Addition of free interactions and mean-field contribution would cause 1