Many of the models for the origin of platinum-group element (PGE) mineralization have stemmed from studies on the Bushveld Complex, especially the Merensky Reef and UG2 Chromitite. All explain certain aspects of the mineralization and associated geology, but can be shown to be inadequate in respect to other features. The model of PGE association with sulphide can be questioned as it predicts that the footwall and hanging-wall sequences to mineralized rocks ought to have formed from sulphide-saturated magmas and should contain more sulphide (and Cu and Ni) than actually observed. This model also requires that the parental magmas to the Lower and Critical Zones contained far more sulphide than is inferred from the observed mineralogy and chemistry of these sequences. The abundance of PGE relative to sulphide (and Cu) associated with the UG2 Chromitite is difficult to reconcile with this model, as it requires extremely high partition coefficients for PGE between sulphide and silicate liquids. The extent of oxidative removal of sulphide from chromite-rich horizons can be shown, from the low concentrations of Cu in these rocks, to have been limited. The association of PGE with chromite is very strong. However, the chemical or physical influence of chromite as a collector of PGE cannot be the only process, as the Platreef contains almost no chromite. Upward migration of late-stage fluids, scavenging PGE from the footwall, and their deposition in mineralized layers fails to explain the observed Cu:Ni:S ratios and abundances in the footwall sequences. The ratio of pyrrhotite:pentlandite:chalcopyrite in the Merensky Reef is comparable to that predicted for an immiscible sulphide liquid, rather than that produced by a later fluid. Also, in terms of this model, the vertical separation between mineralized layers ought to have had an influence on the grade of PGE mineralization in the overlying layers. However, despite enormous variations in thickness between the UG2 and Merensky Reef in different parts of the Bushveld Complex, near-constant grades in the latter are reported. Perhaps the strongest argument against upward migration of PGE is the fact that the Platreef occurs along the basal contact with no igneous footwall cumulates from which to scavenge PGE. A reassessment of the base metal to PGE ratios (exemplified by Cu/Pd ratios) of all rocks in the Lower and Critical Zones indicates that they may be explained as a consequence of accumulation of a PGE phase, which is not sulphide, and may be a PGE-rich alloy phase. The stability of PGE nuclei or clusters in silicate magma has been debated on the basis of experimental studies, which are themselves controversial. Results are equivocal. However, even if such phases do form directly from the magma, the processes of accumulation have yet to be rigorously examined. The basal chromitites of most of the cyclic units in the Critical Zone, not just the UG2 and Merensky, are anomalously enriched in PGE. It is suggested that each cycle overlies an unconformity, possibly resulting from magma addition during which there is a period of non-deposition or even erosion. Such periods of non-accumulation of silicate minerals may be important for the accumulation of the PGE, but again the mechanism has yet to be identified. Whereas there may be more than one process which actually concentrates PGE in the ores of the Bushveld Complex, it appears that the various well-defined models, which are sufficiently specific that they can be quantitatively testable, have deficiencies which question their applicability. Other hypotheses are not yet adequately constrained as to be rigorously testable.