Lead and multiple sulphur isotope compositions were measured in-situ by SIMS on sulphide minerals from phoscorites and carbonatites of the ca. 2.06 Ga Phalaborwa Complex in South Africa. Additionally, sulphide mineral separates and bulk-rock samples were analyzed with IRMS methods to confirm SIMS data. Lead isotope ratios define a trend stretching from unradiogenic to highly radiogenic ratios corresponding to a Pb–Pb regression date of 2054 ± 99 Ma. This apparent date is consistent with the timing of emplacement and thus provides an age estimate for the sulphide mineralization. The least radiogenic Pb isotope compositions overlap, and the regression line intersects, a hypothetical mixing line between MORB mantle and an upper crustal reservoir at ca. 2.1 Ga, suggesting that either a significant quantity of crustal Pb contributed to sulphide mineralization, or that sulphidic xenomelts were derived from an isotopically enriched mantle source. Sulphur isotope ratios of individual sulphide minerals obtained by SIMS are highly variable (δ34S: −15 to +15‰ V-CDT) and, importantly, reveal the contribution of pre-Great Oxidation Event (GOE) atmospheric sulphur with mass-independent isotope fractionation (Δ33S = δ33S–[(1+δ34S)0.515-1]×1000 ≠0.0‰). Mass-independent sulphur isotope fractionation is also revealed by sulphur isotope ratios measured on sulphide mineral separates (Δ33S: 0.2 to 0.7‰) and bulk rock samples (Δ33S: 0.2 to 0.4‰). Generally, the range of sulphur isotope ratios obtained with SIMS is much larger than that observed in non-SIMS data, possibly reflecting isotopic variability at the μm scale, resolvable only with microbeam measurements. Various sources and mechanisms by which supracrustal material may have been incorporated into mantle-derived carbonatite-phoscorite magmas are assessed, taking into account that geological evidence for the presence of sedimentary material available for assimilation during shallow-level magma emplacement is lacking. Given the variability in S and Pb isotopic compositions, it is inferred that pre-GOE surficial Pb and S were not derived from asthenospheric mantle contaminated with supracrustal materials. Instead, whole rock trace element compositions, in concert with published geochemical and petrological evidence, are consistent with interaction of asthenospheric, plume-derived melt with compositionally heterogeneous lithospheric mantle that was metasomatically modified by fluids and melts released from a subducting slab. Despite geochemical and geochronological similarities with the 2055 Ma Busvheld Complex, lead and sulphur isotope data for both complexes are resolvably different, pointing to distinct lithospheric mantle sources involved in sulphide mineralization.