Abstract Sex differences in habitat use (habitat segregation) are widespread in sexually dimorphic ungulates. A possible cause is that males are more sensitive to weather than females, leading to sex differences in sheltering behaviour (the ‘weather sensitivity hypothesis’). However, this hypothesis has never been tested. We considered the allometric rates of net energy gain during times of cold weather and food shortage in a model. We argue that the higher absolute heat losses relative to intake rates of larger ungulates should indeed lead to higher weather sensitivity in males than in females. Furthermore, we tested the weather sensitivity hypothesis empirically in red deer, Cervus elaphus, on the Isle of Rum, U.K. We predicted that (1) use of relatively exposed, high-quality forage habitat should be negatively influenced by bad weather; and (2) this influence should be stronger in males. We found that bad weather (strong wind, low temperature, heavy rain) in winter and spring influenced use of high-quality forage habitat negatively in all deer; that adult males responded more strongly to low temperature and strong wind than did females; and that adult males foraged on windy days at better sheltered sites than did females. Thus, the weather sensitivity hypothesis is supported both theoretically and empirically. We suggest that the weather sensitivity hypothesis can potentially explain winter habitat segregation in a large number of ungulate species.