Elucidating the origin, composition and physical properties of primary kimberlite melts is crucial to our understanding of their source, petrogenesis, ascent mechanisms and ultimately the origin of diamonds. Recently, there has been a growing interest in the study of olivine, which is one of the most abundant minerals in kimberlites with xenocrystic, metamorphic (mantle) and magmatic origins. Olivine is one of the earliest minerals to crystallise in kimberlite magmas, and the presence of ubiquitous zoning (e.g. cores, internal zones, transitional zones, rims, rinds, outmost rinds) and different generations (i.e. primary, pseudosecondary and secondary) of crystal/melt/fluid inclusions in euhedral olivine grains has been shown to provide fundamental insights into the composition and evolution of kimberlite melts. In this contribution, we review and evaluate the following: (1) the widely accepted notion that kimberlite olivine has two distinct origins—xenocrystic and magmatic. We present detailed electron microprobeX-ray element maps of well-preserved and zoned euhedral olivine microcrysts from the Koala and Mark (Lac de Gras, Canada) and Udachnaya-East (Siberia, Russia) kimberlites to show that the cores of olivine occasionally adopt euhedral shapes, which is commonly defined by the distribution of Ni. We present a scenario in which mantle olivine was recrystallised by the early (or proto-) kimberlite melt/fluid infiltrating through the lithospheric mantle to form euhedral ‘pyrocrysts’ (i.e. olivine that formed via re-crystallisation in the mantle in the presence of a melt), which in turn become cores for the subsequent crystallisation of magmatic olivine during kimberlite magma ascent and emplacement. (2) The evolution of ideas using different geochemical, petrological, experimental and melt inclusion approaches to constrain the composition of the primary/parental kimberlite melt. Based on our assessment of available data, in particular using melt inclusions, we propose that kimberlites originated from melts that were initially Si-poor, and Na-K-F-Cl-P-S-bearing and Ca-Mg-carbonate-rich. With this model composition for the primary/parental kimberlite melt considered, we emphasise the implications for the evolution of olivine and its role in the kimberlite petrogenesis. Furthermore, we present a comprehensive model outlining the key stages involved in the petrogenesis of kimberlites, ranging from the generation of the proto kimberlite melt in the mantle, its interaction with mantle silicates during ascent, the role of liquid immiscibility in driving magma differentiation and CO2 degassing and its emplacement and modification in the crust. Finally, we discuss prospective directions that may further guide the future of kimberlite petrological research.