Description: We report new major, trace and volatile element contents (H2O, CO2, F, S, Cl), and new Sr, Nd, Pb and He isotopes on submarine glasses from the Galapagos Archipelago from several dredging expeditions. Four groups are distinguishable on the basis of composition and geographical distribution: the Fernandina group (3He/4He 〉 22 RA), which is similar to the less degassed primitive mantle; the Sierra Negra group (enriched Pb and Sr isotopes, 3He/4He = 8–20 RA), produced by mixing the Floreana (HIMU-type) and Fernandina end-members; the Pinta group (high Δ7/4, Δ8/4 and Th/La ratios, 3He/4He = 6–9 RA), an enriched mantle (EM)-type mantle indicative of recycled material in the source; and the depleted mantle (DM) group, characterized by an isotopic composition similar to mid-ocean ridge basalts (MORB). Only a single submarine glass with the isotopic composition of the Floreana end-member has been identified in the sample suite. Degassing has significantly lowered the glass CO2 content with little effect on the H2O concentration. Volatile data for oceanic basalts reveal that CO2–H2O gas–melt equilibration at eruption depth is common in ocean island basalts (OIB) and rare in MORB, suggesting different ratios of melt transport to bubble formation and gas–melt equilibration. The Galapagos glasses range from sulfide saturated to undersaturated, and a subset of samples indicate that S degasses at pressures ≤ 400 bars. Assimilation of hydrothermally altered material affected the volatile contents of a number of samples in the groups. Once shallow-level processes have been accounted for, we evaluate the volatile contents in the different Galapagos mantle sources. Ratios between volatile and refractory elements with similar incompatibilities are used to estimate the volatile budget of the Galapagos mantle plume. Most of the glasses from the Fernandina, Sierra Negra and Pinta groups have high volatile/refractory element ratios, whereas a few pristine DM group lavas have ratios similar to those measured in MORB. The volatile/refractory element ratios are consistent with previous reports for the high 3He/4He, HIMU and MORB components. The values measured for the Pinta group, however, are higher than those found in other OIB associated with the presence of recycled material (EM-type). Our data suggest that mixing between the different mantle components is pervasive throughout the archipelago, which acts to normalize the volatile data between the groups. The Fernandina component can be modeled by a 6–20% mixture of the high 3He/4He primitive mantle component with the MORB source, assuming a two-layered mantle and using existing estimates of helium concentrations. The resulting estimated volatile content and H/C mass ratio for the high 3He/4He primitive mantle are consistent with previous estimates, but calculated C/3He ratios are lower than the canonical ratio. This indicates the following: (1) the estimates require ∼20–50 times higher C or lower 3He contents, which is difficult to reconcile with the measured volatile/refractory ratios in oceanic basalts; (2) the C/3He ratio is not constant throughout the mantle; (3) an impact erosion model, rather than a two-layered mantle model, is more consistent with the relatively constant C/3He ratios observed in oceanic basalts, although it is unclear how representative oceanic basalts are of the lower mantle. The high volatile content of the high 3He/4He component will affect mantle dynamics and melt migration during plume–ridge interaction as this component would be predicted to be less viscous than the ambient mantle. The lower viscosity material would have an enhanced vertical upwelling, which could explain the buoyancy flux of the Galapagos plume without the need for a temperature anomaly. A lower viscosity, high 3He/4He component could also provide an explanation for the lack of high 3He/4He in Galapagos Spreading Center lavas erupting in the vicinity of the Galapagos plume.