Numerical biophysical models are a standard methodology used to provide estimates of larval dispersal and population connectivity for marine species with a bi-phasic life cycle. However, confidence on biophysical models, tested by confronting model estimates with empirical evidence, is seldomly assessed. We compared time-series estimates of larval supply of the stalked barnacle Pollicipes pollicipes to rocky shores in 3 regions around the Iberian Peninsula (Asturias, Galicia, SW Portugal) with recruitment observations made at monthly intervals for 2 yr. Supply estimates were made with the Regional Ocean Modelling system using several larval behaviour scenarios, while the number of recruits on the stalks of adult barnacles was used as a measure of recruitment intensity. Cross-correlation analysis showed that passive, surface-dwelling and ontogenetically migrating larval scenarios generally produced significant positive correlations at time lags of 0 to +2 mo at the regional level but not at sub-regional or site levels. None of the scenarios produced a substantially better fit than the others, and all 3 produced estimates of average realized dispersal (-73 to +63 km in the S/W and N/E directions) and larval retention (2.0 to 2.4%) that were numerically very similar. These estimates indicate high levels of connectivity, either during larval life or via steppingstone processes, within and between the 3 regions. Based on these estimates, we advocate that the management of the P. pollicipes fishery requires an interactive, polycentric governance system at transregional, regional and local scales, which will give the resource a higher possibility of persistence by diversifying the fishery’s management portfolio.