(1) Counts through time were compiled for five sea otter (Enhydra lutris) populations in the north-east Pacific Ocean that were below equilibrium density: Attu Island, south-east Alaska, British Columbia, Washington State, and central California. Similar data were obtained from the equilibrium density population at Amchitka Island in 1971 and 1986.

(2) Shorelines of Attu and Amchitka islands each were divided into forty-five segments, within which lineal (length of shore at mean higher high water) and areal (mean higher high water to the 10-fathom (18.3-m) depth contour) measures were made of the amount of habitat.

(3) Rate of increase for the four northern populations was 17-20% year-1. Density- or size-dependent changes in rate of increase could not be demonstrated for any of these populations. The California population, in contrast, has undergone three apparent growth phases: the early 1900s to the mid-1970s when it increased about 5% year-1; the mid-1970s to the mid-1980s when it declined about 5% year-1; and the mid-1980s to 1988 when it increased about 7% year-1. An exponential growth model accounted for 92-98% of the variation in counts through time in all cases.

(4) Population increase at Attu Island was achieved largely by range expansion as opposed to increased density. Range expansion in lineal and areal habitat occurred at 11% and 13% year-1, respectively; neither rate was lower (P > 0.25) than the observed rate of increase in numbers of animals counted.

(5) Despite similarities in island size and physical environment, the most conservative estimates of population density at Amchitka Island were > 3 X greater than maximum density estimates for Attu Island.

(6) Surveys of Amchitka Island from the mid-1930s through the mid-1980s indicate that the population increased to a peak in the 1940s; declined abruptly thereafter; and subsequently increased to a new and higher equilibrium in the 1960s, where it has since remained.

(7) These population data, together with information on sea otter foraging and benthic community structure at Attu and Amchitka islands, suggest that multiple population equilibria exist in this system, emanating from complex trophic interactions low in the food web. I hypothesize that the lower population equilibrium is achieved largely or exclusively on an invertebrate diet consisting principally of herbivorous sea urchins. When unregulated by sea otter predation, the rocky benthos is deforested by sea urchin grazing. As growing otter populations compete increasingly for food, grazing intensity declines and the system shifts to one dominated by kelp beds, in turn leading to increased production, a shift in habitat structure, and population increases of kelp bed fishes. Apparently this new food resource elevates the sea otter population to a higher and more stable equilibrium.