Abstract. We describe a relativistic approach to the calculation of nucleon-nucleon Bremsstrahlung, where all meson-baryon and meson-baryon-photon interactions can be calculated consistently and microscopically. In this first relativistic approach to the problem, we present numerical results including both single-scatter and rescatter contributions with a relativistic current density, within a model where the explicit photon coupling to meson-exchange currents are small. The need for high precision (p, PY) measurements is stressed. PACS: 13.75.Cs; 25.10.+s; 27.30.+y The consistent microscopic prediction of (p, PT) observables requires the calculation of the photon coupling to an interacting meson-baryon system. For processes where the photon couples to a free nucleon before or after strong-interactions, one can express the (P, PT) invariant amplitude in the two-potential formalism of GellMann and Goldberger [1], consisting of a one-body nuclear current density and an off-shell NN T-matrix. The essential advantage of this approach is that the NN Tmatrix is defined to all orders in the strong-interaction coupling constant. All recent (p, p 7) calculations use this two-potential formalism, and obtain the NN T-matrices from the Lippmann-Schwinger [24] or BlankenbeclerSugar [5,6] equations for a non-relativistic boson-exchange potential such as Bonn [7] or Paris [8], or from inverse-scattering methods [9], and therefore include only the positive frequency components of the two-nucleon wavefunctions. Under this restriction, only the positive frequency processes that are shown in Fig. 1 are included.In all existing (p, P7) calculations, where the non-relativistic strong-interaction models neglect the negativefrequency components of the off-shell nulceons, it is necThis work is supported by COSY-KFA Jfilich (41140512) and Deutsche Forschungsgemeinschaft (Ga 153/11-4) essary to apply some kind of non-relativistic reduction to the current operator. This may take the form [2, 41 of a Foldy Wouthuysen transformation to obtain a nonrelativistic current with 'relativistic' spin corrections, which are retained to some approximate order, according to where the infinite p/m-expansion is truncated. In a more approximate scheme, it may take the form [6] of a direct Pauli reduction. These two approaches lead to differences of order 7% in the (p, pT) cross section at 280 MeV [10], and differ from the results that are obtained with a completely non-relativistic current density [3,5] by as much as 15% [-2, 6]. The corresponding differences in the spin-observables are generally even larger.This suggests the need for relativistic wave functions in (p, PT) calculations, so that the relativistic current density can be calculated without recourse to any (p/m)_ expansion, and the off-shell two-nucleon state can be properly described as including both its the positive and negative frequency contributions, as shown in Fig. 1. In the Feynman-Stfikelberg interpretation, the negative frequency contributions represent NN creation and annihilat...