Typically, in bag-stack or silo fumigations the concentration of phosphine is not constant, and yet most of what is known about phosphine efficacy against grain insects comes frorn studies with fixed concentrations. Indeed, where changing concentration experiments have been performed, researchers have been unable to explain observed efficacy on the basis of data frorn fixed concentrations. The ability to predict insect mortality in relation to changing phosphine concentrations would facilitate the development of effective fumigation protocols. In this paper, we explore the prospects for making such predictions. After reviewing published and new results, we conclude that the commonly used concentration x time (Ct) product is y e l iable for this purpose. New results, for a strongly resistant strain of Rhyzopertha dominica frorn Australia, suggest that the relationship Cnt = kmay be useful for predicting mortality of this type of insect in changing concentrations. However, in the case of a strain of Sitophilus oryzae with a type of resistance common in Australian S. oryzae, the relationship C t = k proved to be less reliable.