In radar altimetry the electromagnetic (EM) bias is originated by the smaller reflectivity of wave crests than troughs, thus the average sea surface height is under-estimated. Bias uncertainty is currently the largest factor in altimetry error budgets. The EM bias in a bistatic forward-scattering configuration at L-band, such as in Global Navigation Satellite SystemsReflectometry (GNSS-R) altimetry, remains one of the major sources of uncertainty in the altimetry error budget. In this work the EM bias is computed using numerical simulations.To do so, a time-dependent synthetic non-Gaussian sea surface is created using the PiersonMoskowitz and Elfouhaily sea surface height spectra and spreading function. The sea surface is then discretized in facets and "illuminated" using a Right Hand Circular Polarization data at C-and Ku-bands. Then, the numerical model is applied at L-band, for bistatic configurations, including different azimuth angles, and different wind speeds. It is found that the EM bias is almost insensitive to the sea surface spectra selected and increases with increasing wind speed and incidence/scattering angle (up to ~20 cm at θi,s = 45° and U10 = 12 m/s), and it also exhibits a non-negligible azimuthal dependence, that must be accounted for in the error budgets of upcoming GNSS-R altimetry missions.