At mid-shore levels on rocky shores in New South Wales, grazing gastropods are the dominant species in sheltered areas. Where wave-exposure is great, barnacles occupy most of the space. At intermediate levels of waveexposure, there are mixtures of grazing gastropods and barnacles, and the patterns of occupancy of space, and structure of the community change from time to time. The major species found in these areas are the coronuloid barnacle Tesseropora rosea, the patellid limpet Cellana tramoserica, the smaller acmaeid limpet, Patelloida latistrigata, which is mostly confined amongst barnacles, and the predatory whelk Morula marginalba. The roles of each of these species in determining the structure and persistence of intertidal communities were investigated by experimental manipulations of the densities of each of these organisms. In most experiments, a range of densities of limpets and barnacles was used, rather than the simple removal of all of one species.Recruitment of Cellana was negatively associated with increasing density of adult limpets, and with increasing density of barnacles. Growth and survival of juvenile Cellana were decreased by increasing densities of barnacles, probably because barnacles occupied space, preventing limpets from grazing. Morula had no effect on recruitment or survival of juvenile Cellana. Recruitment of juvenile Patelloida was not affected by different densities of barnacles, but survival to adult sizes was poor in areas where whelks are active. In areas where whelks were removed, Patelloida showed increased survivorship with increased cover of barnacles, probably because Patelloida amongst barnacles found refuge from the superior competitive effects of Cellana.The settlement and subsequent survival of Tesseropora were affected in complex ways by the activities of Cellana. At great densities, Cellana can have deleterious effects on newly-settled barnacles, probably by crushing them whilst grazing. In some areas, and at some densities, however, limpets can have beneficial effects on the recruitment and/or survival of Tesseropora. The limpets graze the juvenile stages of growth of foliose macroalgae, preventing them from growing up to pre-empt the rock-surface (thus preventing settlement of barnacles) or to smother alreadysettled barnacles. The effect of limpets on the recruitment and survival of barnacles in any area is a function of the densities of limpets and barnacles, and the height on the shore and local weather (the latter factors influence the rates of growth of algae).Increased cover of the rock-surface by adult barnacles caused reductions in the densities of Cellana. Limpets migrated away from areas of great cover of barnacles, and, if confined in such areas, starved and lost weight. The dispersion, as well as the density of the barnacles was important in determining the effects of barnacle cover. Where barnacles occupied half the space, but were scattered, leaving only small patches of bare rock, they had the same deleterious effects on Cellana as in areas where they were spread evenly to occupy most of the rock-surface. Thus, barnacles could invade areas dominated by limpets provided they recruited in sufficient numbers. They did not have to saturate an area to displace the Cellana.In these communities, all of the species can be considered to have important roles in the establishment and maintenance of community structure. We conclude that interpretations of the roles of individual species must be based on direct, experimental investigation. In this system, there was no indication that many of the species were functionally insignificant.The present experiments also revealed that the interactions among even a few species are very varied and complex; proper investigation requires considerable replication and repeated experimentation in different places and years.Finally, although the present studies allow reliable interpretations of observed patterns of occupancy and dominance on natural shores, the experiments did not provide a predictive framework to anticipate the future events in any area. This is because of great variability in the timing and intensity of recruitment of planktonic propagules of all the components in the system, and in the density and activity of predatory whelks in different areas. These results suggest that tightly co-evolved community relationships are not likely to be important, even if they appear to exist, in communities where most of the species have widely dispersed pelagic offspring and interact in diverse and complex ways at different densities.