Conservation and management of sea otters can benefit from managing the magnitude and sex composition of human related mortality, including harvesting within sustainable levels. Using age and sex-specific reproduction and survival rates from field studies, we created matrix population models representing sea otter populations with growth rates of 1.005, 1.072, and 1.145, corresponding to stable, moderate, and rapid rates of change. In each modeled population, we incrementally imposed additional annual mortality over a 20-year period and calculated average annual rates of change (lambda). Additional mortality was applied to (1) males only, (2) at a 1:1 ratio of male to female, and (3) at a 3:1 ratio of male to female. Dependent pups (age 0-0.5) were excluded from the mortality. Maintaining a stable or slightly increasing population was largely dependent on (1) the magnitude of additional mortality, (2) the underlying rate of change in the population during the period of additional mortality, and (3) the extent that females were included in the additional mortality (due to a polygnous reproductive system where one male may breed with more than one female). In stable populations, additional mortality as high as 2.4 percent was sustainable if limited to males only, but was reduced to 1.2 percent when males and females were removed at ratios of 3:1 or 0.5 percent at ratios of 1:1. In moderate growth populations, additional mortality of 9.8 percent (male-only) and 15.0 percent (3:1 male to female) maximized the sustainable mortality about 3-10 ten-fold over the stable population levels. However, if additional mortality consists of males and females at equal proportions, the sustainable rate is 7.7 percent. In rapid growth populations, maximum sustainable levels of mortality as high as 27.3 percent were achieved when the ratio of additional mortality was 3:1 male to female. Although male-only mortality maximized annual harvest in stable populations, high male biased mortality in all simulations eventually led to low proportions of males, leading to instability in projected populations over time. Our findings identify the critical need to understand underlying rates of change that can be acquired only through frequent monitoring of managed populations. Models could be improved through better understanding of the effects of density and demographic and environmental stochasticity on sea otter vital rates. Although our primary objective was to provide information useful in managing harvests of sea otters, our findings have implications for the conservation and management of sea otter populations subjected to other sources of mortality that can be quantified, such as incidental, accidental, or illegal.