The 660-km seismic discontinuity, which is a significant structure in the Earth's mantle, is generally interpreted as the post-spinel transition, as indicated by the decomposition of ringwoodite to bridgmanite + ferropericlase. All precise high-pressure and high-temperature experiments nevertheless report 0.5-2 GPa lower transition pressures than those expected at the discontinuity depth (i.e. 23.4 GPa). These results are inconsistent with the post-spinel transition hypothesis and, therefore, do not support widely accepted models of mantle composition such as the pyrolite and CI chondrite models. Here, we present new experimental data showing post-spinel transition pressures in complete agreement with the 660-km discontinuity depth obtained by high-resolution in situ X-ray diffraction in a large-volume high-pressure apparatus with a tightly controlled sample pressure. These data affirm the applicability of the prevailing mantle models. We infer that the apparently lower pressures reported by previous studies are experimental artefacts due to the pressure drop upon heating. The present results indicate the necessity of reinvestigating the position of mantle mineral phase boundaries previously obtained by in situ X-ray diffraction in high-pressure-temperature apparatuses.