Because workers in the eusocial Hymenoptera are more closely related to sisters than to brothers, theory predicts that natural selection should act on them to bias (change) sex allocation to favor reproductive females over males. However, selection should also act on queens to prevent worker bias. We use a simulation approach to analyze the coevolution of this conflict in colonies with single, once-mated queens. We assume that queens bias the primary (egg) sex ratio and workers bias the secondary (adult) sex ratio, both at some cost to colony productivity. Workers can bias either by eliminating males or by directly increasing female caste determination. Although variation among colonies in kin structure is absent, simulations often result in bimodal (split) colony sex ratios. This occurs because of the evolution of two alternative queen or two alternative worker biasing strategies, one that biases strongly and another that does not bias at all. Alternative strategies evolve because the mechanisms of biasing result in accelerating benefits per unit cost with increasing bias, resulting in greater fitness for strategies that bias more and bias less than the population equilibrium. Strategies biasing more gain from increased biasing efficiency whereas strategies biasing less gain from decreased biasing cost. Our study predicts that whether queens or workers evolve alternative strategies depends upon the mechanisms that workers use to bias the sex ratio, the relative cost of queen and worker biasing, and the rates at which queen and worker strategies evolve. Our study also predicts that population and colony level sex allocation, as well as colony productivity, will differ diagnostically according to whether queens or workers evolve alternative biasing strategies and according to what mechanism workers use to bias sex allocation.