Thermodynamic modeling alone cannot uniquely characterize the evolution path (T-P-t path) of a magmatic system, because the assumption of reaction reversibility precludes recovery of the history of that evolution from the final state that is preserved in the rock record. Nevertheless, as a tool for discriminating possible scenarios for magma evolution, such models are informative. This point is illustrated with two examples. The first involves modeling equilibrium and fractional crystallization of midocean ridge basalts. Attention is focused on the variation of intensive parameters (like oxygen fugacity) and heat output with crystallization, and how the latter can impact fluid dynamical modeling. The second example illustrates modeling mantle metling during closed-system adiabatic ascent. Here thermodynamic modeling leads to quantitative predictions of melt productivity, which may be used to interpret melt generation rates and chemical characteristics of lavas produced from volcanic centers.