Temporal trends in the recovery of exploited species in marine protected areas (MPAs) are useful for a proper assessment of the efficacy of protection measures. The effects of protection on the fish assemblages of the sublittoral rocky reefs in the "Penisola del Sinis-Isola di Mal di Ventre" MPA (W. Sardinia, Italy) were evaluated using a multi-year series of data. Four surveys, conducted 7, 10, 13 and 15 years after the area was designated as an MPA and carried out in the period spanning June and July, were used to estimate the abundance and biomass of commercial species. The surveys were carried out in zones with decreasing levels of fishing restrictions within the MPA (zones A, B, C) and in unprotected zones (OUT1 and OUT2), and underwater video visual census techniques were used. Protected zones only occasionally showed higher levels of abundance or biomass, and the trajectories of those metrics were not consistent across the years. In addition, the zone with the highest level of protection (zone A) never presented levels of abundance and biomass higher than those in zones B and C. This study shows that even 15 years after designation, protection has had no appreciable effect in the MPA studied. It is argued that this is emblematic of several shortcomings in the planning, regulation and enforcement frameworks of the MPA.