Quantum electrodynamics (QED) theory predicts that vacuum under the influence of a strong magnetic field is birefringence. Recently, several groups have proposed to used a high finesse Fabry—Perot cavity to increase the average path length of the light in the magnetic field. This together with the state-of-the-art dipole magnets, should bring the effect within reach of observation. However, the mirrors used in the FP are known to have intrinsic birefringence which is of orders of magnitude larger than the birefringence of the vacuum. In this letter, we analyze the effect of uncontrollable variations of mirror birefringence on two recently proposed optical schemes. The first scheme,1 which we called the frequency scheme, is based on measurement of the beat frequency of two orthogonal polarized laser beams in the cavity. We show that mirror birefringence contributes to the detection uncertainties in first order, resulting in a high susceptibility to variations of its value. In the second scheme, which we called the polarization scheme, laser polarized at 45° relative to the B-field is injected into the cavity. The ellipticity and polarization rotation of the light exiting the cavity is measured.2 Under this scheme, mirror birefringence contributes as a correction of the QED effect, greatly reducing its sensitivity to the undesirable changes.