The face-centered cubic phase π(Fe,Co,Ni,S) has been shown to exist, at room temperature, within wide composition limits in or close to the M9S8section of the quaternary system Fe–Co–Ni–S. The M:S ratio of the binary phase π (Co,S) is 9:8 with very narrow homogeneity ranges on both sides of Co9S8, but in π (Fe,Co,Ni,S) the ratio is somewhat higher and appears to increase with decreasing cobalt content. Stoichiometric Co9S8probably contains a small number of vacancies in both sublattices. It is quite lilcely that the sulphur sublattice is nearly fully occupied and that departures from stoichiometry are caused by the varying degree of occupancy of the metal sublattice.The crystal structure, which was proposed for Co9S8and for the mineral pentlandite by Lindqvist etal., has been confirmed for these two substances and for π (Fe,Co,Ni,S) in general by X-ray and neutron powder diffraction. The present evidence does not support the crystal structure suggested for natural pentlandite by Eliseev; Eliseev's model does not, in fact, account for the diffraction data of any of the substances examined in this work.Replacement of cobalt in π (Co,S) by iron or nickel or both results in an expansion of the unit cell, the maximum increase in a(π) amounting to about 3%. Cobalt in π (Co,S) cannot be replaced completely by iron or by nickel in samples prepared by dry synthesis, but if the substitution is simultaneous, the π structure will be preserved over a considerable range of compositions even on total replacement. The stability limits of π (Fe,Ni,S) have been found somewhat wider than those stated by Lundqvist.In π phases with the compositions Co8MS8the metal atoms can conceivably be present in ordered sublattices. This possibility was explored by neutron diffraction in slowly cooled Co8NiS8. Unlike in spinels, where nickel shows a strong preference for octahedral co-ordination, the cobalt and nickel atoms were found to be distributed at random.