I studied the effects of microclimate on waterfowl energetics at a willow roost site on the floodplain of the Mississippi River. This study was an attempt to determine the functional value of willows to waterfowl on a Midwestern wetland complex. Specifically, I conducted a comparative study of the importance of habitat structure to the thermal energetics of male mallards (Anas platyrhynchos) during migration and winter. I tested the hypotheses that microclimate was thermally superior in willows than in adjacent habitats and that male mallards roosting in willows experienced greater thermoregulatory energy savings. On clear days, shortwave radiation exceeded 600 W/m$\sp2$ in the deep water because of reflection off the water surface. There were no significant differences in air temperature or relative humidity among habitats. Wind speed was lowest in willows (typically less than 1.0 m/s), intermediate in the deep water, and highest in corn when wind was from the north or northwest. No significant differences were detected in operative temperature of male mallards among habitats at night. Both air and operative temperatures typically differed by less than 2-3 C among habitats. Despite the dense canopy associated with willows, net radiation at night only differed by 30 W/m$\sp2$. Operative temperatures differed significantly among habitats during the day, especially in spring when canopy cover and stem density were nil in the deep water and corn. Shortwave radiation was 250-300 W/m$\sp2$ lower in willows than in the other habitats and operative temperature was up to 13 C lower in willows. Costs of thermoregulation for male mallards were lowest in willows from 1600-0800. However, thermoregulatory costs were higher in willows than in other habitats during the day. Male mallards expended 15% less energy in willows than in corn on cold, clear, windy nights. Diurnal energy savings were greatest in deep water when wind was from the north or west, but thermal superiority shifted to the corn when wind was from the east. Mallards had a lower critical temperature of 3-12 C and combined with their large body size, were probably not thermally stressed until operative temperatures dropped near freezing. Because mallards require open water, they emigrate when ice covers much of the available habitat and therefore reduce their exposure to thermally stressful conditions.