The role played by the urchins, Paracentrotus lividus and Arbacia lixula, in the formation and persistence of barren areas dominated by encrusting coralline macroalgae is yet to be fully elucidated. This study, carried out in the NW Mediterranean (43° 30′N, 10° 20′E) between February 2005 and April 2006, investigated how the loss or density decrease in one or both urchin species influences the recovery of erect macroalgal stands (dominated by filamentous forms) at the margins of barren areas. At a depth of 4–6 m, three barren patches were assigned to each of the following treatments: (1) control (natural densities of A. lixula and P. lividus); (2) 50 % of the natural density of A. lixula and natural density of P. lividus; (3) total removal of A. lixula and natural density of P. lividus; (4) 50 % of the natural density of P. lividus and natural density of A. lixula; (5) total removal of P. lividus and natural density of A. lixula; (6) 50 % of the natural densities of both A. lixula and P. lividus; (7) total removal of both A. lixula and P. lividus. The effects of the herbivore treatments were evaluated either in the presence or the absence of encrusting corallines. The partial or total removal of A. lixula, P. lividus or both favored the proliferation of filamentous macroalgae at the margins of barren patches. The presence of encrusting corallines reduced the development of these macroalgae. The results of this study suggest that a moderate decrease in the density of just one of the two species can decrease the ability of the herbivore assemblage to control the proliferation of filamentous macroalgae at the margins of barren patches. The extent of barren areas appears, therefore, to be regulated by the outcome of density-dependent interactions between the two species of sea urchins.