Geochemical fronts for Sr isotopes have been identified within carbonate mylonites along the Gavarnie Thrust in the central Pyrenees. Over 200 samples have been analyzed for Sr-87/Sr-86, including samples from seven measured sections through the thrust zone. A geochemical profile has been drawn through the maximum Sr-87/Sr-86 values at each sample site, and two separate geochemical fronts identified. The data show that Sr was transported southward along a 2m thick mylonite zone for at least 1.5 km. Fluid inclusion crush-leach analyses show that a Sr-rich hypersaline brine with a high Sr-87/Sr-86 ratio was trapped within Triassic redbeds beneath the Thrust; this is a likely candidate for the fluid that passed through the mylonites, increasing their Sr-87/Sr-86 ratios from depositional values close to that of Cretaceous seawater. Comparison of the Sr content of fluid inclusions and calcite in the same vein systems allows the solid-fluid partition coefficient, K(v-Sr) to be estimated. Values range from 0.5 in the northern part of the thrust system to 1.9 farther south, where fluids are more dilute. These values can be combined with minimum transport distances for the geochemical fronts to give minimum time-integrated fluxes of between 1800 and 3000 m3/m2. The estimates of K(v-Sr) are supported by partial-derivativeO-18 data on carbonates, which show a good correlation with Sr-87/Sr-86 in the southern part of the thrust system but a wide range in values at high Sr-87/Sr-86 in the north. This is expected if K(v-Sr) was similar to K(v-O) in the south but lower in the north. Veins in the carbonates almost invariably have more radiogenic Sr than the adjacent matrix. This suggests a highly organized flow pattern, with unidirectional flow parallel to the mylonitic banding. Veins in hangingwall phyllites have similar Sr-87/Sr-86 ratios to the mylonitized Cretaceous limestones and are less radiogenic than the adjacent matrix, suggesting limited escape of fluid upward out of the carbonates into hangingwall vein systems. Although the precise shapes of geochemical profiles cannot be constrained, Damkohler numbers were probably in the range 20 to 100, indicating relatively rapid rates of fluid-rock equilibration compared with fluid flow velocity. Comparison of flux estimates suggests that time-averaged permeabilities were three orders of magnitude higher in the mylonites than in undeformed rocks and varied by a factor of about two within the mylonites. Overall, fluid flow can be explained by a fault-valve model, with slow pervasive flow through the mylonite during plastic deformation punctuated by rapid flow into veins during seismic events.