We present a calibration for attenuated total reflectance (ATR) micro-FTIR for analysis of H2O in hydrous glass. A Ge ATR accessory was used to measure evanescent wave absorption by H₂O within hydrous rhyolite and other standards. Absorbance at 3450 cm⁻¹ (representing total H₂O or H₂O_t) and 1630 cm⁻¹ (molecular H₂O or H₂O_m) showed high correlation with measured H₂O in the glasses as determined by transmission FTIR spectroscopy and manometry. For rhyolite, wt%H₂O=245(±9)×A₃₄₅₀-0.22(±0.03) and wt%H₂O_m=235(±11)×A₁₆₃₀-0.20(±0.03) where A₃₄₅₀ and A₁₆₃₀ represent the ATR absorption at the relevant infrared wavelengths. The calibration permits determination of volatiles in singly polished glass samples with spot size down to ~5 μm (for H₂O-rich samples) and detection limits of ~0.1 wt% H₂O. Basaltic, basaltic andesite and dacitic glasses of known H₂O concentrations fall along a density-adjusted calibration, indicating that ATR is relatively insensitive to glass composition, at least for calc-alkaline glasses. The following equation allows quantification of H₂O in silicate glasses that range in composition from basalt to rhyolite: wt%H₂O=(ω×A₃₄₅₀/ρ)+b where ω = 550 ± 21, b = −0.19 ± 0.03, ρ = density, in g/cm³, and A₃₄₅₀ is the ATR absorbance at 3450 cm⁻¹. The ATR micro-FTIR technique is less sensitive than transmission FTIR, but requires only a singly polished sample for quantitative results, thus minimizing time for sample preparation. Compared with specular reflectance, it is more sensitive and better suited for imaging of H₂O variations in heterogeneous samples such as melt inclusions. One drawback is that the technique can damage fragile samples and we therefore recommend mounting of unknowns in epoxy prior to polishing. Our calibration should hold for any Ge ATR crystals with the same incident angle (31°). Use of a different crystal type or geometry would require measurement of several H₂O-bearing standards to provide a crystal-specific calibration.