Abstract Forty-two volcanic rocks of the Panjal Traps were analysed for platinum-group elements (PGEs) to investigate the magma genesis, high-temperature behaviour and exploration potential of these elements. The PGE data exhibit substantial variability and show no systematic relation to their low- or high-Ti affinity. Instead, the basalts can be subdivided into a PGE-undepleted group (group 1) that has ΣPGE >10 ppb and Cu/Pd <30 000, and a PGE-depleted group, which consists of a subgroup showing limited (group 2A) or substantial depletion in Ir-series PGEs relative to Ni (group 2B). The group 1 samples indicate an S-undersaturated history, whereas the group 2 samples might have different origins in terms of S-saturation. Fractionation of a tiny amount of sulfide melts (0.075–0.1%) from a representative group 1 sample accounts for the chalcophile element patterns observed in the group 2B samples. The relatively high Cu/Pd, unfractionated Ni/Ir and low PGE abundances observed in the group 2A samples cannot be explained by equilibration of an immiscible sulfide melt alone, and probably require decomposition of residual sulfides into sulfide melt and a monosulfide solid solution in the mantle restite. Our results question the notion that the coexistence of PGE-undepleted and -depleted magmas is prospective in the exploration of magmatic Ni–Cu–(PGE) sulfide mineralization.