Interactions between sea otters Enhydra lutris, sea urchins Strongylocentrotus droebachiensis, and kelp were investigated following the reduction in sea otter density in Prince William Sound, Alaska, after the Exxon Valdez oil spill in 1989. At northern Knight Island, a heavily oiled portion of the sound, sea otter abundance was reduced by a minimum of 50 % by the oil spill, and from 1995 through 1998 remained at an estimated 66 % lower than in 1973. Where sea otter densities were reduced, there were proportionally more large sea urchins. However, except in some widely scattered aggregations, both density and biomass of sea urchins were similar in an area of reduced sea otter density compared with an area where sea otters remained about 10 times more abundant. Furthermore, there was no change in kelp abundance in the area of reduced sea otter density. This is in contrast to greatly increased biomass of sea urchins and greatly reduced kelp density observed following an approximate 90 % decline in sea otter abundance in the western Aleutian Islands. The variation in community response to a reduction in sea otters may be related to the magnitude of the reduction and the non-linear response by sea urchins to changes in predator abundance. The number of surviving sea otters may have been high enough to suppress sea urchin populations in Prince William Sound, but not in the Aleutians. Alternatively, differences in response may have been due to differences in the frequency or magnitude of sea urchin recruitment. Densities of small sea urchins were much higher in the Aleutian system even prior to the reduction in sea otters, suggesting a higher rate of recruitment.