Früh-Green et al. (1996), we calculate serpentinization temperatures ranging between 430° and 590°C. However, a sensitivity analysis shows that large errors affect these temperatures (about ±100°C). Nevertheless, we think we can determine that the serpentinization occurred at temperatures higher than 350°C.Using the mass-balance equation developed by Taylor (1977), and assuming that seawater (δ 18 θ = Woo) or 18 O hydrothermally enriched seawater (δ 18 θ = +2.4%c) is the serpentinizing fluid, we can determine that the wateπrock ratios (in oxygen atom proportions) were, at 400°C, around 0.55 ± 0.1 and 1.05 ± 0.2, respectively, and at 500°C, 0.4 ± 0.1 and 0.8 ±0.1, respectively.The relatively high serpentinization temperatures derived from this study imply that serpentinization has occurred at significant depth in the axial lithosphere. This has potentially important consequences, which we discuss, on the amount of tectonic rotation during the uplift of the ultramafic rocks to the seafloor, on the orientation of principal stresses at depth beneath the axial valley, and on the seismic velocity and density structure of the oceanic lithosphere.All types of serpentine (meshwork and vein) from Site 920 appear to have formed at these relatively high temperatures. However, previous oxygen-isotope studies Hebert et al., 1990) of cores from Site 670 reported ultramafic rocks that were serpentinized at lower temperature (<200°C). This indicates that the ultramafic component of the axial lithosphere in the area did not become entirely serpentinized at depths corresponding to rock temperatures of at least 350°C.