High-selenium copper refinery anode slimes form two separate and dynamically evolving series of compounds with increasing electrolysis time. In one, silver is progressively added to non-stoichiometric copper selenides, both those originally present in the anode and those formed subsequently in the slime layer, and in the other, silver-poor copper selenides undergo a dis-continuous crystallographic sequence of anodic-oxidative transformations. The silver-to-selenium molar ratio in the as-cast anode and the current density of electrorefining can be used to construct predominance diagrams for both series and, thus, to predict the final bulk “mineralogy” of the slimes. Although totally incorrect in detail, these bulk data are sufficiently accurate to provide explanations for several processing problems which have been experienced by Kidd Creek Division, Falconbridge Ltd., in its commercial tankhouse. They form the basis for a computer model which predicts final cathode quality from chemical analyses of smelter feed.