AbstractWe assessed by numerical modeling the coastal fish larval dispersion along the southern coast of Mallorca (Balearic Islands, NW Mediterranean) with the objective of determining the factors that contribute to successful recruitment. We assumed that fish larvae dispersal is mainly regulated by physical transport. Currents are mainly wind driven in this area; therefore, changes in wind forcing have a first‐order impact on larval transport. The synoptic wind patterns were systematically analyzed based on self‐organizing map analysis. The wind fields were clustered using a neural network pattern recognition approach into two modes, producing opposite along‐shelf flow. The seasonal changes between spring and summer in the dominance of either mode modulate the along‐shelf circulation, producing flow shifts under some circumstances. This variability in the wind regime was consistent throughout the 10 years analyzed (2000–2009). Using the Princeton Ocean Model (POM) and a particle‐tracking algorithm, we analyzed the effect of wind‐forced currents in the connectivity among near‐shore habitats. We show that, at the spatial scale considered, the coastal morphology and stochastic wind forcing favor local recruitment (mean of 30% self‐recruitment). Maximum transport distances of 20–30 km were typically associated with particles left to drift for 21 days. The implications for the performance of the four marine protected areas near SW Mallorca Island are discussed. Our results suggest that, although wind episodes determine the fate of short‐time spawning events, on a seasonal basis, regular larval supply to coastal zones is ensured by wind stochasticity.