ABSTRACT. We present a mathematical model of larval dispersal into a coastal zone dominated by wlnd-driven currents: larvae are considered as active particles tracked by the 3-D (3-dimensional) velocities calculated by a physical model sophisticated enough to give a good representation of the 3-D structure of wind-driven currents. As a n application, we model the larval recruitment of the annelid Owenia fusiformis in Banyuls Bay, France, located at the southwest extremity of the Gulf of Lions. The currents are generated by the permanent offshore Liguro-Provencal circulation and by time-varying winds, the s\vunmlng behav~our of larvae is assumed to be only vertical, the spawning zones are in 2 adjacent bays and the pelagic phase lasts about 4 wk. Larvae are assumed to be successfully recruited if they settle after pelagic morphological changes on the substratum suitable to their benthic development. Larvae which do not belong to the successful recruitment either settle too young on the suitable substratum or go out of the suitable substratum. The model shows that the fate of the larvae tracked by the currents appears to b e determined early in the pelagic phase by being trapped in low current zones. Simulations allow the deduction of the advection losses for different winds the proportion of larvae reaching the adult habitat is 60% at maximum for downwelling conditions and 15 O/o at minimum for init~a l u p w e h n g conditions. Our results indicate too that under the most frequent wind cond~tions no mixing occurs between populations of the 2 adult habitat zones which are 1.6 km apart from each other and that no successful recruitment occurs from larvae coming from bays located at the north of the considered coastal zone. Only combinations of changing wind conditions could permit arrival of larvae from Paulilles Bay to Banyuls Bay. The interest of such individual-based models consists not only in increasIng our understanding of the link between spatial and temporal dynamics of meroplanktonic populations but also in allowing us to explore the potential effects of habitat alterat~on on those populations.