When using the standard beam-propagation method (BPM), we can analyze optical waveguide structures in a clear way. Attention must be paid, however, to the following restrictions of the applicability of the BPM: no large transverse index steps, no large propagation steps compared with the wavelength, and propagation under paraxial conditions. Another problem arises from the finiteness of the computational window. The field intensity should be equal to zero, and radiation modes should not be reflected at the window edges. To achieve these conditions, we usually place artificial absorbers at the edges. This leads to additional storage and computing time.