Some low-chloride pore waters observed in accretionary complexes are thought to result from clay dehydration and subsequent migration of the released water along faults or sand layers. We test this hypothesis with a two-dimensional flow and transport model for a cross section of the northern Barbados accretionary complex. The model flow system is driven by consolidation of the accreted sediments and by fluids from smectite clay dehydration. Steady state simulations result in concentrations that are too high along the d¿collement fault and too low near the seafloor. In a transient model we simulate buildup and release of fluids by assuming that strain or hydrofracture along the fault causes an instantaneous increase in d¿collement permeability of 2--3 orders of magnitude. With such an increase, the observed concentrations can be achieved in 100--1000 years. Also pressures along the fault rise to near lithostatic values in 10--100 years and remain high for 1000--10,000 years. This pressure rise may represent a mechanism for sustaining high fault permeabilities long after the initial increase. ¿ American Geophysical Union 1995