<p>Granitic melts derived from anatexis of metasedimentary rocks, particularly biotite-dehydration reactions, are important lithium (Li) sources. Petrogenetic models depend on mineral-melt lithium partitioning, yet published partition coefficients (<InlineEquation ID="IEq1"> <EquationSource Format="TEX">\({D}_{{{{\rm{Li}}}}}^{{{{\rm{mineral/melt}}}}}\)</EquationSource> <EquationSource Format="MATHML"><math> <msubsup> <mrow> <mi>D</mi> </mrow> <mrow> <mi mathvariant="normal">Li</mi> </mrow> <mrow> <mi mathvariant="normal">mineral/melt</mi> </mrow> </msubsup> </math></EquationSource> </InlineEquation>) vary by over an order of magnitude, and are commonly used as static values. Here we use thermodynamic modelling coupled with relevant published <InlineEquation ID="IEq2"> <EquationSource Format="TEX">\({D}_{{{{\rm{Li}}}}}^{{{{\rm{mineral/melt}}}}}\)</EquationSource> <EquationSource Format="MATHML"><math> <msubsup> <mrow> <mi>D</mi> </mrow> <mrow> <mi mathvariant="normal">Li</mi> </mrow> <mrow> <mi mathvariant="normal">mineral/melt</mi> </mrow> </msubsup> </math></EquationSource> </InlineEquation> ranges, including a dynamic composition- and temperature-dependent <InlineEquation ID="IEq3"> <EquationSource Format="TEX">\({D}_{{{{\rm{Li}}}}}^{{{{\rm{biotite/melt}}}}}\)</EquationSource> <EquationSource Format="MATHML"><math> <msubsup> <mrow> <mi>D</mi> </mrow> <mrow> <mi mathvariant="normal">Li</mi> </mrow> <mrow> <mi mathvariant="normal">biotite/melt</mi> </mrow> </msubsup> </math></EquationSource> </InlineEquation>, to quantify viable enrichment during partial melting and fractional crystallisation. Using the lithium-rich Cornubian granite batholith, we demonstrate the sensitivity of results to <InlineEquation ID="IEq4"> <EquationSource Format="TEX">\({D}_{{{{\rm{Li}}}}}^{{{{\rm{mineral/melt}}}}}\)</EquationSource> <EquationSource Format="MATHML"><math> <msubsup> <mrow> <mi>D</mi> </mrow> <mrow> <mi mathvariant="normal">Li</mi> </mrow> <mrow> <mi mathvariant="normal">mineral/melt</mi> </mrow> </msubsup> </math></EquationSource> </InlineEquation> choices for modally-dominant lithium-poor phases (e.g. quartz), as well as phases traditionally thought to dominate lithium budgets (e.g. biotite). While economic lithium enrichment can result from extreme fractionation with selective <InlineEquation ID="IEq5"> <EquationSource Format="TEX">\({D}_{{{{\rm{Li}}}}}^{{{{\rm{mineral/melt}}}}}\)</EquationSource> <EquationSource Format="MATHML"><math> <msubsup> <mrow> <mi>D</mi> </mrow> <mrow> <mi mathvariant="normal">Li</mi> </mrow> <mrow> <mi mathvariant="normal">mineral/melt</mi> </mrow> </msubsup> </math></EquationSource> </InlineEquation>, we suggest that dehydration melting of fluorinated biotite is the most viable petrogenetic model. The latter reconciles the common observations of fluorite in lithium-granites and their late-orogenic occurrence, and provides a mechanism for extensive fractionation.</p>