The sulfur contents and sulfur isotopic compositions of 24 glassy submarine volcanics from the Mariana Island Arc and back-arc Mariana Trough were determined in order to investigate the hypothesis that subducted seawater sulfur (δ 34S= 21‰) is recycled through arc volcanism. Our results for sulfur are similar to those for subaerial arc volcanics: Mariana Arc glasses are enriched in 34S (δ 34S=up to10.3‰,mean= 3.8‰) and depleted in S (20–290 ppm, mean = 100 ppm) relative to MORB (850 ppm S,δ 34S= 0.1 ± 0.5‰). The back-arc trough basalts contain 200–930 ppm S and haveδ 34S values of1.1 ± 0.5‰, which overlap those for the arc and MORB. The low sulfur contents of the arc and some of the trough glasses are attributed to (1) early loss of small amounts of sulfur through separation of immiscible sulfide and (2) later vapor-melt equilibrium control of sulfur contents and loss of sulfur in a vapor phase from sulfide-undersaturated melts near the minimum in sulfur solubility atƒO 2 ≈NNO (nickel-nickel oxide). Although these processes removed sulfur from the melts their effects on the sulfur isotopic compositions of the melts were minimal. Positive trends ofδ 34S with 87Sr/ 86Sr, LILE and LREE contents of the arc volcanics are consistent with a metasomatic seawater sulfur component in the depleted sub-arc mantle source. The lack of a 34S-rich slab signature in the trough lavas may be attributed to equilibration of metasomatic fluid with mantle material along the longer pathway from the slab to the source of the trough volcanics. Sulfur is likely to have been transported into the mantle wedge by metasomatic fluid derived from subducted sediments and pore fluids. Gases extracted from vesicles in arc and back-arc samples are predominantly H 2O, with minor CO 2 and traces of H 2S and SO 2. CO 2 in the arc and back-arc rocks hasδ 13C values of −2.1 to −13.1‰, similar to MORB. These data suggest that degassing of CO 2 could explain the slightly lowerδ 13C values for some Mariana Trough volcanic glasses, and that incorporation of subduction-derived organic carbon into the Mariana Trough mantle source may not be necessary. More analyses are required to resolve this question, however.