Despite the abundance of SO2(g) in magmatic gases, precursory increases in magmatic SO2(g) are not always observed prior to volcanic eruption, probably because many terrestrial volcanoes contain abundant groundwater or surface water that scrubs magmatic gases until a dry pathway to the atmosphere is established. To better understand scrubbing and its implications for volcano monitoring, we model thermochemically the reaction of magmatic gases with water. First, we inject a 915??C magmatic gas from Merapi volcano into 25??C air-saturated water (ASW) over a wide range of gas/water mass ratios from 0.0002 to 100 and at a total pressure of 0.1 MPa. Then we model closed-system cooling of the magmatic gas, magmatic gas-ASW mixing at 5.0 MPa, runs with varied temperature and composition of the ASW, a case with a wide range of magmatic-gas compositions, and a reaction of a magmatic gas-ASW mixture with rock. The modeling predicts gas and water compositions, and, in one case, alteration assemblages for a wide range of scrubbing conditions; these results can be compared directly with samples from degassing volcanoes. The modeling suggests that CO2(g) is the main species to monitor when scrubbing exists; another candidate is H2S(g), but it can be affected by reactions with aqueous ferrous iron. In contrast, scrubbing by water will prevent significant SO2(g) and most HCl(g) emissions until dry pathways are established, except for moderate HCl(g) degassing from pH < 0.5 hydrothermal waters. Furthermore, it appears that scrubbing will prevent much, if any, SO2(g) degassing from long-resident boiling hydrothermal systems. Several processes can also decrease or increase H2(g) emissions during scrubbing making H2(g) a poor choice to detect changes in magma degassing. We applied the model results to interpret field observations and emission rate data from four eruptions: (1) Crater Peak on Mount Spurr (1992) where, except for a short post-eruptive period, scrubbing appears to have drastically diminished pre-, inter-, and post-eruptive SO2(g) emissions, but had much less impact on CO2(g) emissions. (2) Mount St. Helens where scrubbing of SO2(g) was important prior to and three weeks after the 18 May 1980 eruption. Scrubbing was also active during a period of unrest in the summer of 1998. (3) Mount Pinatubo where early drying out prevented SO2(g) scrubbing before the climactic 15 June 1991 eruption. (4) The ongoing eruption at Popocate??petl in an arid region of Mexico where there is little evidence of scrubbing. In most eruptive cycles, the impact of scrubbing will be greater during pre- and post-eruptive periods than during the main eruptive and intense passive degassing stages. Therefore, we recommend monitoring the following gases: CO2(g) and H2S(g) in precursory stages; CO2(g), H2S(g), SO2(g), HCl(g), and HF(g) in eruptive and intense passive degassing stages; and CO2(g) and H2S(g) again in the declining stages. CO2(g) is clearly the main candidate for early emission rate monitoring, although significant early increases in the intensity and geographic distribution of H2S(g) emissions should be taken as an important sign of volcanic unrest and a potential precursor. Owing to the difficulty of extracting SO2(g) from hydrothermal waters, the emergence of > 100 t/d (tons per day) of SO2(g) in addition to CO2(g) and H2S(g) should be taken as a criterion of magma intrusion. Finally, the modeling suggests that the interpretation of gas-ratio data requires a case-by-case evaluation since ratio changes can often be produced by several mechanisms; nevertheless, several gas ratios may provide useful indices for monitoring the drying out of gas pathways. Published by Elsevier Science B.V.