Abstract <p>Two experimental approaches are used to estimate the conditions of kimberlite melt generation. (1)&#xa0;Investigation of partial melting of carbonated ultramafic rocks. (2) Determination of the conditions of the stability of polymineralic assemblages on the kimberlite liquidus. The analysis of available experimental data and estimates of the compositions of primary kimberlite melts showed that the residual mineral assemblage during kimberlite generation included, probably, olivine, low-Ca pyroxene, and garnet. The approximation of experimental data on the compositions of melts equilibrated with such an association allowed us to calculate pressure in the melt generation zone and CO₂ content in the primary melt. Using the obtained relationships, equilibrium parameters were calculated for the supposed primary kimberlite melts, and it was argued that they could be produced by partial melting of carbonated peridotite at pressures of 40–90 kbar and temperatures of 1600–1800°C. The real temperatures could be significantly lower because of the presence of water and other volatiles. The calculations demonstrated that the primary melts contained 10–20 wt % CO₂ and were not saturated with respect to high-Ca pyroxene and magnesite. The retention of magnesite in the residue requires an increase in melt CO₂ content up to ~30 wt %, whereas the equilibrium with olivine and low-Ca pyroxene imposes high SiO₂ content (~15 wt %) and high temperature of deep melts, much above a normal cratonic geotherm.</p>