Five compositionally distinct suites of mafic dykes, ranging from quartz and olivine tholeiites to slightly alkaline dolerites and picrite-ankaramites, were emplaced in the Bunger Hills ∼ 1140 Ma ago, shortly after high-grade metamorphism and intrusion of mafic to felsic plutonic rocks. Less abundant alkali basalts and trachybasalts were emplaced 500 Ma ago. The geochemistry of most of these suits is complex, in spite of their restricted geographic distribution, and implies mechanisms such as dynamic partial melting, RTF magma chambers, and various depths of magma segregation. Some degree of crustal contamination (by a siliceous, LILE- and 87Sr-rich partial melt or fluid) in AFC magma chambers is also possible, but isotopic and geochemical data are equally consistent with, and indeed require, variable LILE and LREE metasomatism of the mantle source regions. Apart from major differences in incompatible element (LILE, LREE, HFSE) ratios between suites, which imply derivation from at least six distinct mantle source reservoirs, there are significant intra-suite variations which also imply source heterogeneity (both lateral and vertical). The source reservoirs range from slightly depleted (Sr i 0.7029, ϵ Nd + 6.3, low K/Zr, Ba/Zr, etc.) for one group of transitional olivine dolerites, through unenriched to moderately enriched (Sr i 0.7032–0.7051, ϵ Nd −0.7 to + 3.9) for other dolerites, to variably metasomatised, long-term enriched (Sr i 0.7074–0.7168, ϵ Nd −18.6 to −17.2) for alkali basalts. At least three source components were involved in generation of the dykes, comprising depleted (asthenospheric?) mantle, Nb-poor enriched (lithospheric) mantle (probably including subducted sedimentary material), and a Nb-rich OIB-type component. Metasomatism of the source of the alkali basalts during the late Archaean or early Proterozoic may have been contemporaneous with continental crust formation in the area.