Temperature and H2O activity can be determined with high precision using metamorphic mineral assemblages that define both a dehydration equilibrium and a temperature-sensitive cation-exchange equilibrium. Such determinations are obtained by applying the Gibbs method and then integrating two resulting differential equations, as illustrated here for the assemblage garnet-chlorite-quartz. The first equation, a geothermometer that monitors temperature based upon Fe−Mg exchange between garnet and chlorite, was calibrated using rocks at Pecos Baldy, New Mexico: 0=0.05 P(bars)−19.02 T(K)+4607 ln K D+24,156 with errors of ±8°C based upon analytical precision. The second equation monitors differences in the activity of water between specimens (1) and (2): 0=(0.1 X Mg−chl, 1 − 2.05)(P 2 − P 1) +[−33.02+5.96 ln(X Fe−chl, 1/X alm, 1)][T 2−T 1 −2.67 RT 1ln[a(H2O)2/a(H2O)1] +5.96 T 1ln(X Fe−chl, 2 X alm, 1/X Fe−chl, 1 X alm, 2). For samples equilibrated at the same pressure and temperature, microprobe analytical errors of 1% limit precision to ±0.01 a(H2O). For samples equilibrated at the same pressure but variable temperature, uncertainty of ±8°C limits precision to ±0.06 a(H2O). Extreme presure sensitivity requires that the H2O-barometer be applied only to rocks where pressure gradients are absent or well-constrained. The geothermometer gives temperatures in agreement with two other garnet-chlorite geothermometers (Dickenson and Hewitt 1986; Ghent et al. 1987) and with garnet-biotite geothermometry (ferry and Spear 1978) over the temperature range 350–520°C. Application of the relative H2O barometer shows variations in the activity of water approaching 0.30 in several study areas. Either pelitic schists commonly equilibrate with a fluid that is not pure H2O, or some pelitic rocks undergo metamorphism in the absence of a free fluid phase.