Aim: We evaluated whether patterns of species diversity (α, β and γ) of rocky shore assemblages followed latitudinal gradients (i.e. LDGs) along the South American coasts, and tested hypotheses related to potential processes sustaining or disrupting the expected LDG pattern at various spatial scales. Location: Coasts of South America. Taxon: Macroalgae and sessile/slow-moving macrofauna on intertidal rocky shores. Methods: We evaluated changes in species composition across 143 sites. The degree of replacement and loss of species at different spatial scales (i.e. coasts, regions and sites) were estimated to help distinguish among ecological, historical and evolutionary hypotheses for explaining LDGs. Furthermore, components of diversity and taxonomic distinctness were measured, and variability in these measures was decomposed using analysis of covariance. Finally, we examined relationships between diversity and a suite of environmental and anthropogenic variables to identify potential mechanisms that may be responsible for the reported spatial relationships. Results: Species composition varied with latitude, and this variability was relatively consistent on both coasts. At all spatial scales, replacement of species was the dominant phenomenon (>95%), rather than loss in the total number of species (<5%). LDGs were strongly dependent on the diversity component and the spatial scale: generally, positive for regional β-diversity, negative for α-diversity and site β-diversity. Sea surface temperature (SST) was the variable that best explained patterns of diversity along both coasts (14%–22%), but other regional and local environmental variables associated with river discharges, upwelling, confluence of currents, tides and anthropogenic pressures also accounted for an important portion of variation (5%–14% each). Main conclusions: Species diversity of South American rocky shores followed, with interruptions, LDGs. The trend of those LDGs, however, depended on the scale and metric used to describe diversity. It is proposed that patterns of LDGs at various scales are not the result of a single overarching process but are strongly influenced by local and regional processes. Although the most evident environmental gradient was the decrease in SST towards the south, it was demonstrated that regional and local environmental variables were also important for understanding the increase in regional β-diversity towards the tropics.