Djerfisherite (K6(Fe,Ni,Cu)25S26Cl) occurs as an accessory phase in the groundmass of many kimberlites, kimberlite-hosted mantle xenoliths, and as a daughter inclusion phase in diamonds and kimberlitic minerals. Djerfisherite typically occurs as replacement of pre-existing Fe–Ni–Cu sulphides (i.e. pyrrhotite, pentlandite and chalcopyrite), but can also occur as individual grains, or as poikilitic phase in the groundmass of kimberlites. In this study, we present new constraints on the origin and genesis of djerfisherite in kimberlites and their entrained xenoliths. Djerfisherite has extremely heterogeneous compositions in terms of Fe, Ni and Cu ratios. However, there appears to be no distinct compositional range of djerfisherite indicative of a particular setting (i.e. kimberlites, xenoliths or diamonds), rather this compositional diversity reflects the composition of the host kimberlite melt and/or interacting metasomatic medium. In addition, djerfisherite may contain K and Cl contents less than the ideal formula unit. Raman spectroscopy and electron backscatter diffraction (EBSD) revealed that these K–Cl poor sulphides still maintain the same djerfisherite crystal structure. Two potential mechanisms for djerfisherite formation are considered: (1) replacement of pre-existing Fe–Ni–Cu sulphides by djerfisherite, which is attributed to precursor sulphides reacting with metasomatic K–Cl bearing melts/fluids in the mantle or the transporting kimberlite melt; (2) direct crystallisation of djerfisherite from the kimberlite melt in groundmass or due to kimberlite melt infiltration into xenoliths. The occurrence of djerfisherite in kimberlites and its mantle cargo from localities worldwide provides strong evidence that the metasomatising/infiltrating kimberlite melt/fluid was enriched in K and Cl. We suggest that kimberlites originated from melts that were more enriched in alkalis and halogens relative to their whole-rock compositions.