Deep fluids are important for the evolution and properties of the lower crust and upper mantle in tectonically active settings. Uncertainty about their chemistry has led past workers to use upper crustal fluids as analogues. However, recent results show that fluids at >15 km differ fundamentally from shallow fluids and help explain high-pressure metasomatism and resistivity patterns. Deep fluids are comprised of four components: H2O, non-polar gases (chiefly CO2), salts (mostly alkali chlorides), and rock-derived solutes (dominated by aluminosilicates and related components). The first three generally define the solvent properties of the fluid, and models must account for observations that H2O activity may be quite low. The contrasting behavior of H2O-gas and H2O-salt mixtures yields immiscibility in the ternary system, which can lead to separation of two phases with fundamentally different chemical and transport properties. Thermodynamic modeling of equilibrium between rocks and H2O using simple ionic species known from shallow-crustal systems yields solutions possessing total dissolved solids and ionic strength that are too low to be consistent with experiments and resistivity surveys. Addition of CO2 further lowers bulk solubility and conductivity. Therefore, additional species must be present in H2O, and H2O-salt solutions likely explain much of the evidence for fluid action in high-P settings. At low salinity, H2O-rich fluids are powerful solvents for aluminosilicate rock components that are dissolved as previously unrecognized polymerized clusters. Experiments show that, near H2O-saturated melting, Al-Si polymers comprise >80% of solutes. The stability of these species facilitates critical critical mixing in rock-H2O systems. Addition of salt (e.g., NaCl) changes solubility patterns, but aluminosilicate contents remain high. Thermodynamic models indicate that the ionic strength of fluids with Xsalt = 0.05 to 0.4 and equilibrated with model crustal rocks have predicted bulk conductivities of 10-1.5 to 100 S/m at porosity of 0.001. Such fluids are thus consistent with conductivity anomalies commonly observed in the lower crust (e.g., the “G” anomaly), and are capable of the mass transfer commonly seen in metamorphic rocks exhumed from the lower crust and subduction zones.