The iron oxidation state in silicate melts is important for understanding their physical properties, although it is most often used to estimate the oxygen fugacity of magmatic systems. Often high spatial resolution analyses are required, yet the available techniques, such as μXANES and μMössbauer, require synchrotron access. The flank method is an electron probe technique with the potential to measure Fe oxidation state at high spatial resolution but requires careful method development to reduce errors related to sample damage, especially for hydrous glasses. The intensity ratios derived from measurements on the flanks of FeLα and FeLβ X-rays (FeLβf/FeLαf) over a time interval (time-dependent ratio flank method) can be extrapolated to their initial values at the onset of analysis. We have developed and calibrated this new method using silicate glasses with a wide range of compositions (43–78 wt% SiO2, 0–10 wt% H2O, and 2–18 wt% FeOT, which is all Fe reported as FeO), including 68 glasses with known Fe oxidation state. The Fe oxidation state (Fe2+/FeT) of hydrous (0–4 wt% H2O) basaltic (43–56 wt% SiO2) and peralkaline (70–76 wt% SiO2) glasses with FeOT > 5 wt% can be quantified with a precision of ±0.03 (10 wt% FeOT and 0.5 Fe2+/FeT) and accuracy of ±0.1. We find basaltic and peralkaline glasses each require a different calibration curve and analysis at different spatial resolutions (∼20 and ∼60 μm diameter regions, respectively). A further 49 synthetic glasses were used to investigate the compositional controls on redox changes during electron beam irradiation, where we found that the direction of redox change is sensitive to glass composition. Anhydrous alkali-poor glasses become reduced during analysis, while hydrous and/or alkali-rich glasses become oxidized by the formation of magnetite nanolites identified using Raman spectroscopy. The rate of reduction is controlled by the initial oxidation state, whereas the rate of oxidation is controlled by SiO2, Fe, and H2O content.