We discus existent conflicts between experimentally measured and theoretically calculated melting curves of Mo, Ta, and W. By assuming that vacancy formation plays a fundamental role in the melting process, an explanation for the measured melting curves is provided.Furthermore, we show that the Lindemann law fits well all the measured melting curves of bcc transition metals if the Grüneisen parameter is written as a power series of the interatomic distance. For completeness, we examine possible reasons for current disagreements between shockwave and DAC experiments. To solve them, we propose the existence of an extra high P-T phase for Mo, Ta, and W.In materials science and geophysics, it is essential to understand how elements undergo the transition between solid and liquid phases at high pressures. For example, an accurate picture of materials performance under extreme conditions and detailed models of planetary interiors both depend on this knowledge. In recent years, the generation and measurement of simultaneous high pressures and high temperatures has undergone rapid development with the diamond-anvil cell (DAC) technique [1 -3]. As a consequence of this, the amount of available data on melting behaviour at high pressure is increasing [4,5]. The bcc transition metals (i.e. molybdenum (Mo), tantalum (Ta), and tungsten (W)) are especially interesting in this regard because they have very high melting points at ambient pressure and, in addition, at room temperature (RT) they remain in a stable body-centered cubic (bcc) structure up to extremely high pressures [6,7]. Mo, Ta, and W therefore are valuable test cases for studying how materials melt.Despite a large number of experimental and theoretical efforts, however, agreement on the melting behaviour of Mo, Ta, and W at high pressure has been elusive. A number of experimental studies have been carried out with the use of shock-wave measurements [8 -10] and laser-heated diamond-anvil cells [11,12]. Likewise, theoretical studies have attempted to predict the melting curve of Mo, Ta,. Unfortunately, substantial variation exists in the results of these earlier efforts owing to the use of different experimental and theoretical techniques.The above-mentioned discrepancies are illustrated in Fig. 1 for the case of Ta.This figure presents the melting curves reported by different authors [11,12,13,15,18] including a new data point (3850 ± 150 K at 98 GPa) obtained by us using the technique described in Ref. [12]. It shows that at 30 GPa a melting temperature of 3600 ± 100 K was measured using the in situ speckle method [11] and synchrotron x-ray diffraction [12] whereas theoretical calculations [13] and Lindemann estimates [15] give a melting