Stomach contents of 1032 lobsters Homarus amencanus from an area with dense macroalgae (n = 115) and sea urchin (Strongylocentrotus droebachiensis) denuded barrens (n = 917) off the southwestern coast of Nova Scotia were examined to compare natural dlets. Gross dletary composition and mean stomach fullness appeared similar in these 2 habitats; the diet spectrum encompassed a broad range of plants and animals but was dominated by molluscan, crustacean, echinoderm, and polychaete remains. Mussels (Modiolus rnodiolus, Mytilus edulis) appeared the most consistently important identified prey species, in terms of frequency-of-occurrence and estimated volume (points) indexes. Patterns in prey points were investigated by discriminant analysis. Shifts in point values for gastropods, ascidians, hermit crabs and brachyuran crabs proved useful in discerning habitat. Dietary diversity (mean number of prey taxa types per stomach) was significantly less for lobsters collected on barrens in September 1979 (12.1) than for lobsters taken from the macroalgal area in August 1979 (13.9). Temporal variations in diet were investigated for the barrens. There were significant decreases in mean stomach fullness and mean number of prey taxa types per stomach for lobsters collected in February-March 1982 as opposed to July-August 1981. Concomitantly. temporal shifts in diet composition were evident from discriminant analysis, predominantly through decreased importance of sea urchins, ophiuroids, lobsters, and brachyuran crabs in February-March. Data on the dietary importance of sea urchins in reiaiiu~i iu uiilel p e y i a a did n~t siippor: the *potkcsis tkat !cbs!ers are =. pri!?cinal -,--, 'keystone' predator on sea urchins. There is still insufficient evidence to assess whether lobster production is food limited on barrens.