Drilling in the Cascadia accretionary complex enable us to evaluate the contribution of dehydration reactions and gas hydrate dissociation to pore water freshening. The observed freshening with depth and distance from the prism toe is consistent with enhanced conversion of smectite to illite, driven by increase in temperature and age of accreted sediments. Although they contain gas hydrate -as evidenced by discrete low chloride spikes- the westernmost sites drilled on Hydrate Ridge show no freshening trend with depth. Strontium data reveal that all the m¿lange samples contain deep fluids modified by reaction with the subducting oceanic crust. Thus we infer that, at the westernmost sites, accretion is too recent for the sediments to have undergone significant illitization. Our data demonstrate that a smooth decrease in dissolved chloride with depth cannot generally be used to infer the presence or to estimate the amount of gas hydrate in accretionary margins.