Based on the survival of sea otters held at rehabilitation centers during the 1989 Exxon Valdez oil spill in Alaska, we built two models of otter mortality. One was based on the relationship between mortality and distance from spill origin, the other was based on the relationship between mortality and time from the spill origin. These models are simplistic and are meant as first steps in arriving at realistic risk estimates and in providing a conceptual framework for relating oil spills and sea otter mortality. Using the distance model, we simulated the impact of an Exxon Valdez event occurring at different locations along the California coast. A spill at the Monterey Peninsula had the greatest impact, exposing 90% of the California sea otter population to oil and killing at least 50% of the individuals. The time model was used to predict the mortality of otters exposed to oil of various ages and for various periods of time. It suggested that efforts to rehabilitate otters should be discontinued 20-30 d after a spill. The limitations of the data available from the Exxon Valdez spill emphasize the importance of being prepared to conduct appropriate research during the next oil spill in sea otter habitat.