AbstractForaging involves a trade‐off between food and safety. Most research into the trade‐off invokes safety from predation. But danger and its associated risk arise from multiple causes that cannot be assessed solely with reference to predators. A more complete assessment of risk management requires experimental designs that attempt to modify and measure risks, regardless of the source of danger. I aimed to do so by adding shelter (mulched straw) and time‐varying supplemental food (rabbit chow), while measuring foraging behavior and habitat use by a seminatural population of meadow voles. Voles foraged more intensely under safety, recognized least risk when given access to both food and shelter, but altered their risk management through time: management included a novel form of sex‐dependent habitat selection in which male–male pairs occupied risky areas without shelter while female–female pairs occupied habitats sheltered by straw. The pattern is consistent with a sex‐dependent evolutionary game in which female territoriality and tolerance of other females limit conflict with, and space use by, males. Voles' array of interacting strategies demonstrates that ecologists must be wary of ascribing risk only to predation, and particularly so if experiments are blind to other dangers and processes that alter foraging behavior and habitat selection.