We present a theory based on first-principles calculations explaining (i) why the tunability of spin polarizations of photoelectrons from Bi2Se3 (111) depends on the band index and Bloch wavevector of the surface state and (ii) why such tunability is absent in the case of isosymmetric Au (111). The results provide not only an explanation for the recent, puzzling experimental observations but also a guide toward making highly-tunable spin-polarized electron sources from topological insulators.Since the beginning of spintronics, constant efforts have been made to generate electrons with a high degree of spin polarization using transport [1], optical [2], and magnetic resonance methods [3]. In particular, optical methods, also known as optical spin orientation, use polarized-light irradiation. For example, electrons in the valence band of strained and surface-treated GaAs can be excited by circularly polarized light and emitted with ∼ 80 % spin polarization [4]. GaAs photocathodes are widely used as spin-polarized electron source in lowenergy electron microscopy [5], in accelerators used in high-energy physics [6], etc.Recently, it has been proposed that topological insulators can serve as a spin-polarized electron source when irradiated with polarized light [7]. By changing the polarization of light and the direction toward which photoelectrons are collected, one can obtain an electron beam which is spin-polarized in an arbitrary direction, with a 100 % degree of spin polarization [8] (the measured degree is over 80 % [9]). On the other hand, the direction of spin polarization of electrons generated from a strainedGaAs photocathode is fixed by the surface-normal direction perpendicular to which the strain is applied. Moreover, unlike GaAs photocathodes, in which the photon energy is fixed to ∼ 1.5 eV by the material band gap, photocathodes using a topological insulator can be operated within a wide range of photon energies. Even if there could be several technological hurdles that should be overcome, topological insulators are conceptually new candidates for photocathodes for spintronics.Despite these recent developments, we still do not understand the results from some of the key spin-and angleresolved photoemission spectroscopy (SARPES) experi-