In April of 1987, unusual sea ice severely scoured exposed rocky seashores near Halifax, Nova Scotia. Mid-tide levels of these shores are commonly dominated by a lush canopy of perennial fucoid rockweeds, with an understory of mussels, barnacles and crustose algae. Patches of cleared space in these canopies are often occupied by ephemeral algae. After the scour, emergent rock in mid-shore areas was occupied by a succession of diatoms, green and blue-green ephemeral algae, and then Fucus canopy. Mussels reappeared in the understory, but rarely displaced the rockweed canopy. These events provided an exceptional opportunity to study recovery dynamics in a perennial intertidal community, at a scale (> 40 km) much larger than any in previous studies. This paper describes large-scale patterns of natural succession at a variety of locations and disturbance intensities, and then reviews the community structure in the light of several experimental tests of processes and species interactions during the succession. Formation of a Fucus canopy was ubiquitous, and variations in the recruitment and regeneration of Fucus were major influences on the rate of succession and the abundance of other species. A survey of Fucus and mussel dominance over about 15 km of coastline supports this view. Most areas in this survey were not dominated by mussels, but by the small-scale patch dynamics of ephemeral algae and Fucus, in succession. Previous work suggested that emergent rock on these shores is normally dominated by fucoid rockweeds because predatory whelks control competition with the mussels, and because herbivorous littorinid snails control competition with ephemeral algae. However, our observational and experimental results show that these interactions were not generally important during this succession, and that the importance of different species and interactions varies. Our results emphasize the importance of several properties of the rockweeds, as the most abundant and structurally important species. These conclusions suggest caution in the extrapolation of specific, small-scale experiments to general hypotheses about community structure and recovery from disturbance. (C) 1997 Elsevier Science B.V.