The current theoretical investigations of silicon crystallites are discussed with particular emphasis on porous silicon. First of all various calculations of the energy gap are compared with recent experimental data for crystallites. Then the radiative lifetime is determined in terms of purely electronic transitions. This provides a useful scheme only for small crystallites, since for larger clusters, closer to the bulk situation, phonon-assisted transitions will dominate. The relative importance of these two processes is estimated as well as their cross-over vs. size. The following part is concerned with the effect of defects : dangling bonds at the surface and donor (or acceptor) impurities. Non-radiative capture by neutral dangling bonds dominates, so that the presence of one such dangling bond is likely to kill the luminescence. Charged dangling bonds lead to radiative capture which can explain the IR luminescence. The effect of donor impurities is shown to be completely different than in the bulk. The last part is devoted first to a full calculation of the excitonic exchange splitting, showing that the commonly used simplified model is only valid for strongly asymmetric crystallites. It is then shown that the electron-lattice interaction produces a non-negligible Stokes shift which adds to the effect of the exchange splitting. Finally Auger recombination is calculated and shown to be very efficient, which is in line with experimental evidence. A general discussion is finally given, showing that a consistent picture is beginning to emerge.