The Zagros fold belt results from active collision of the Arabian plate with central Iran, and is characterized by the development of a spectacular >200 km-wide fold-train in its sedimentary cover. Although the architecture of this accretionary prism has been extensively studied because of its important implications for hydrocarbon exploration, aspects such as the kinematics of individual folds and the sequence of fold development remain to be investigated in detail. It is commonly believed that the ongoing deformation through the Zagros belt has led to the south-westward migration of the front of the fold belt. In the south-western Fars province (central Zagros), the most frontal structure is delineated by the Mand anticline, a well-exposed detachment fold on the shore of the Persian Gulf. This near-symmetrical anticline involves relatively competent Phanerozoic sedimentary rocks above a regional décollement in Hormuz salt. In order to document the geometry and kinematics of this fold, we have constructed several balanced cross-sections on the basis of a recently published section constrained by seismic data (Letouzey and Sherkati, 2004). Several solutions to the length versus area restoration problem common to detachment folds are then proposed: fault-related folding, detachment folding with internal deformation, and detachment folding accompanied by the flexure of the flanking synclines below the regional stratigraphic level. On the western limb of the anticline, fluvio-marine terraces, tilted by 1.7 to 4.5°, provide an additional constraint on fold kinematics and suggest that surface deformation is most compatible with a detachment fold, probably associated with synclinal flexure. Applying such a model, as well as new 14C ages for the marine terrace deposits, we calculate tilting rates of 0.04 to 0.05°/kyr, which would be produced by a Late Pleistocene shortening rate (perpendicular to the structure) of 3 to 4 mm/yr. Although this preliminary estimate suffers from relatively large uncertainties, mostly due to the absence of independent dating of the terraces and independent constraints on the folding model, we conclude that shortening across the Mand anticline could absorb 20 to 35% of the 8 mm/yr convergence across the entire Zagros. This result is consistent with a normal forward-propagating deformation sequence in a thin-skinned tectonic regime. It also implies that the sedimentary cover of the frontal Zagros is fully decoupled from the basement, most probably at the level of the Hormuz salt, in contrast to recent models that suggested active deformation of the sedimentary cover to be controlled by thrust faults in the basement.