Concentrations of platinum-group elements and sulphides in the Pyroxenite Marker of the upper Main Zone are variable, but generally low (up to 100 ppb PGE and 0.2 weight % S). The metal patterns may mostly be explained by sulphide segregation from PGE depleted residual Upper Critical Zone magma, but they are inconsistent with sulphide segregation from a replenishing influx of undepleted Critical Zone magma. Instead, we favour a model whereby a relatively cool and dense Main Zone crystal mush intruded the Bushveld chamber during the later stages of the deposition of the Upper Critical Zone and displaced warmer and lighter residual magma depleted in chalcophile metals (Sharpe, 1985). Based on the metal contents and textural evidence such as the occurrence of ophitic textures, we hypothesize that the Pyroxenite Marker formed in response to localized supercooling and the suppression of plagioclase crystallization. The model implies that the layer represents a poor target for PGE mineralization in the upper portions of the Bushveld Complex.