Dissolved H2O in silicate melts and glasses plays a crucial role in volcanic eruptions on terrestrial planets and affects glass properties and magma evolution. In this paper, major progress on several aspects of the H2O-melt (or glass) system is reviewed, consistency among a variety of data is investigated, discrepancies are evaluated, and confusion is clarified. On the infrared measurement of total H2O and species concentrations, calibration for a variety of glasses has been carried out at room temperature. The measurements for H2O in rhyolitic glasses have undergone the most scrutiny, resulting in the realization that absorptivities for the near-infrared bands depend on total H2O content. Although the variation of the absorptivities does not seem to significantly affect the determination of total H2O, it does affect the determination of molecular H2O and OH species concentrations. Calibration of the infrared technique for H2O in rhyolitic glasses still needs much improvement, especially at high total H2O. Furthermore, it is now almost certain that the molar absorptivities also depend on the measurement temperature in in situ studies. Hence it will be necessary to carry out calibrations in situ at high temperatures. On H2O speciation, results from two experimental approaches, the quench technique and the in situ technique, are very different, leading to controversy in our understanding of true speciation. A solution is presented to reconcile this controversy. It is almost certain that the quench technique does not suffer from a quench problem, but interpretation of in situ reuslts suffered from ignoring the dependence of the molar absorptivities on measurement temperature. Accurate calibration at high temperatures is necessary for the quantitative application of the in situ technique to H2O speciation in silicate melts and glasses. On H2O solubility in silicate melts, recent experimental work has significantly expanded the T--P range of solubility measurements, and recent solubility models fill a gap for predicting solubility for a wide range of melt compositions. I present a solubility model for rhyolitic and quasirhyolitic melts over a wide range of T and P (500¿--1350 ¿C, 0--8 kbar) by incorporating the role of speciation. The solubility model is able to recover the experimental solubility data and has extrapolative value, although the partial molar volume of H2O derived from the solubility model differs from that derived from density measurements. On H2O diffusion, recent studies on H2O diffusion in a quasi-rhyolitic melt at 800¿--1200 ¿C, 0.5--5 kbar, and up to 7% total H2O not only provide important new diffusion data, but are also challenging earlier understanding of H2O diffusion based on data in rhyolitic glasses at 400¿--550 ¿C, 1 bar, and 0.2--1.8% total H2O. A comparison between the earlier model and recent data is made. The recent high-temperature diffusivities at total H2O≤2% can be predicted by the earlier model. However, at higher total H2O, the earlier model fails. New work is under way to understand the diffusion mechanisms at high H2O contents. ¿ 1999 American Geophysical Union