In standard fluorescence recovery after photobleaching (FRAP) applications for measuring lateral diffusion rates and adsorption/desorption kinetics of fluorescent molecules at biological or model membranes, irreversible bleaching is induced by a bright excitation flash of at least millisecond time scale. It has been presumed that the bleaching event is of a low probability and the significant bleached population that develops during the flash results from each molecule undergoing thousands of excitation/deexcitation cycles before a bleaching event occurs. In some FRAP experiments, notably polarized FRAP (PFRAP) for measuring molecular rotational diffusion rates, it is desirable to use much shorter (subnanosecond) bleaching pulses. However, subnanosecond pulses are shorter than the fluorescence lifetime, so that any fluorophore will experience at most only one visit to the excited state during the bleaching pulse. If bleaching occurs only by the same processes as in slower FRAP experiments, one would thereby expect only minimal bleaching regardless of the bleach intensity. Moreover, the ability of fast polarized pulses to imprint an anisotropic orientational pattern in the postbleach unbleached fluorophore, an ability essential for PFRAP, is not at all guaranteed, particularly if two-photon processes are involved in high-intensity short bleach pulses. In this study, bleaching depths are measured as a function of subnanosecond pulse intensity on a small labeled protein covalently immobilized on fused silica. We show that bright subnanosecond laser flashes do indeed produce significant bleaching, that both two photon effects and reversible bleaching are involved, and that polarized bleaching does produce an anisotropic orientational pattern of unbleached fluorophore. We also postulate a theoretical molecular state model which semiquantitatively accounts for the experimentally observed dependence of reversible bleaching on bleaching pulse intensity.