Tantalum mineralization is associated with granites and pegmatites that invariably are rich in lithium. This in part is due to the extreme fractionation that is necessary for melts to reach saturation in either a Ta or an Li phase. However, there may also be chemical interactions in melts that explain this association. Experiments, therefore, were conducted to determine the effect of lithium on the solubility of tantalite (MnTa 2 O 6 ) in water-saturated granitic melts. Other important elements in rare metal granites include Nb, Zr, Hf, and W. Consequently the effect of Li on the solubilities of columbite (MnNb 2 O 6 ), zircon (ZrSiO 4 ), hafnon (HfSiO 4 ), and hbnerite (MnWO 4 ) in granitic melts was also investigated. The haplogranitic melt compositions used in these experiments have constant mole percent Si and Al, and the Al/(Li + Na + K) ratio was 1.0 for all experiments, i.e., the lithium content was the only compositional variable. Columbite and tantalite solubilities in the granitic liquids increase by a factor of approximately 2 to 3 with the addition of 2 wt percent Li 2 O to haplogranite at 750 degrees to 1,035 degrees C, 2 kbars, and water-saturated conditions. The solubility data were extrapolated to a lower temperature, more reasonable for columbite-tantalite crystallization, then compared to compositions representative of evolved liquids from which rare metal granites and pegmatites crystallize. It is estimated that at 600 degrees C these melt compositions could be saturated in columbite but not tantalite. However, tantalite saturation is predicted if the melts contained less Li and F. Therefore, the genesis of tantalum mineralization may be explained by Ta being retained in the melt because of high Li-F concentrations. Tantalite crystallization is delayed until an Li + or - F + or - P mineral crystallizes, which lowers tantalite solubility and results in a general association of Ta with Li in mineralized granitic rocks.In contrast to columbite and tantalite, the solubility of wolframite is not affected by the lithium content of the melt (with up to 3.8 wt % Li 2 O in starting glasses) and the solubilities of zircon and hafnon decrease with increasing Li content of the melt. One possible interpretation of the hbnerite solubility data is that the melt contains dominantly W (super 6+) , which behaves as a network former. The decrease in zircon-hafnon solubility with Li may reflect the lower field strength of Zr (super 4+) and Hf (super 4+) , compared to Nb (super 5+) and Ta (super 5+) , i.e., Zr and Hf are less able to compete with Li for nonbridging oxygens, whereas Nb and Ta are more able to compete. It is also significant that the solubility of hafnon is greater than that of zircon, similar to the higher solubility of tantalite compared to columbite. This can explain why Zr/Hf and Nb/Ta ratios both decrease with fractionation.