A hydrous fluid phase is critical in controlling effective stress and fault mechanics, and influencing the mineralogy and strength of materials within fault zones. Oxygen and hydrogen isotope and chemical analysis of wall rock gneiss, pseudotachylyte, and selected minerals in gneiss and pseudotachylyte from the Homestake shear zone was used to assess whether melting occurred in the presence of meteoric water or involved only minor amounts of H2O derived from micas in wall rock gneiss. Bulk pseudotachylyte has slightly lower δ18OSMOW than the whole rock protolith. δD for one bulk pseudotachylyte is essentially identical to biotite in gneiss; δD for two samples is lower by ~20?. Bulk pseudotachylyte has lower SiO2 and K2O, and higher Al2O3, FeO, MgO, CaO, and H2O, than gneiss. The lower SiO2 of pseudotachylyte compared to gneiss is explained by physical segregation of 25 to 72 volume % of quartz clasts from the mobile melt phase. Samples of gneiss and pseudotachylyte define a SiO2-δ18O mixing line between quartz and the most SiO2- and 18O-depleted pseudotachylyte. Physical segregation of quartz (highest oxygen isotope composition in the pseudotachylyte-gneiss system) accounts for the slightly lower oxygen isotope composition of bulk pseudotachylyte relative to gneiss. The similar δD of pseudotachylyte and biotite from gneiss in one sample is consistent with dehydration melting of biotite during frictional heating and dissolution of biotite-derived H2O in the melt. Late devitrification of glass and formation of greater amounts of fine-grained muscovite, accompanied by 10--30% loss of hydrogen as H2O, results in lower δD values in other samples. In general, melt generation occurred in a fault zone closed to infiltration of meteoric water. There was no free, H2O-rich pore fluid present at the time of slip to potentially influence the behavior of the fault.