The tendency of powder mixtures to demix due to segregation is a substantial technical concern for manufacturing of pharmaceutical solid dosage forms. In this paper, a methodology for quantifying the maximum potential segregation, as well as predicting the actual, segregation of a powder blend is outlined. The method begins with measurement of the sieve assay profile of the blend. Individual component and overall size segregation indices are defined in terms of overlap integral calculations that characterize the maldistribution of the components among the sieve fractions. This approach has been in use since 2001 at Merck & Co. for multicomponent polydisperse pharmaceutical powder blends and is appropriate for direct compression blends and agglomerated systems, including ordered mixtures. Die-filling performance can be predicted when the individual maldistribution index (Di) is combined with blend flowability (gravity-normalized cohesivity) on a regime map. This map strongly confirms that better blend flowability increases segregation realization. The approach permits rational risk assessment for existing product changes and optimization of new formulations. Analysis assumptions are discussed and deemed to be conservative with respect to prediction of demixing. (C) 2015 Elsevier B.V. All rights reserved.