Abstract The crystal structure of labradorite is characterized by two modulation waves with largely different periods of about 1500 Å and 50 Å, giving rise to s- and e-satellites, respectively. The problems to determine the true structure by diffraction methods are discussed. Since each s-lamella has its own e-modulation, common experiments are not sufficient. Using only s-satellites around main reflections assumes e-averaged structures of the s-lamellae. On the other hand, using only averaged e-satellites and main reflections integrated over the s-satellites leads to a hypothetical, idealised structure such as if there were no s-modulation. For this reason the analysis of an extensive data set, including e-satellites up to third order, cannot resolve the Ca/Na modulation unambiguously, although the Si/Al modulation can be determined consistently from directly refined occupancies and T-O bond lengths.