Fish assemblages associated with shallow (4–7 m deep) artificial rocky habitats (i.e. breakwaters) have been assessed between July 2002 and September 2003, at the marine protected area of Miramare and adjacent areas outside the reserve (northern Adriatic sea). Our purpose was to: (1) detect possible differences between ‘protected vs fished’ breakwaters; and (2) compare two visual-census techniques for fish assessment (i.e. strip transects vs stationary points). The fish assemblages observed between protected and fished breakwaters during all four sampling periods were statistically different. More fish taxa were found at the protected than fished breakwaters, while there was no difference in total fish density. Most fish species targeted by fisheries had a greater density (e.g. Sciaena umbra, Dicentrarchus labrax, Sparus aurata, Diplodus vulgaris, Diplodus sargus and Diplodus puntazzo) and/or size (e.g. S. aurata and D. annularis) at the protected than fished breakwaters. There was a significant difference in fish assemblages due to assessment method. In general, the number of taxa was greater when assessed by strip transects than stationary points. Total fish density was almost unaffected by the method used, while total density of demersal fish (i.e. excluding schooling species) tended to be greater when evaluated by strip transects, although the difference was statistically significant only in one sampling period out of four. These results indicate that protection from fishing may have the potential to influence fish assemblages associated with breakwaters. Additionally, caution should be used when comparing fish assemblage data collected by different visual assessment techniques.