Abstract. Heterothermy plays an important role in lowering the costs of thermoregulation in endotherms by reducing water and energy requirements. We tested predictions that birds in arid habitats should express fine-scale variation in their thermoregulatory patterns as a function of prevailing climatic conditions. We assessed effects of air temperature (T air ) and water vapor pressure deficit (D) on body temperature (T b ) in free-living White-browed Sparrow-Weavers (Plocepasser mahali ) during summer in two arid habitats in the Kalahari Desert, South Africa, using data from a dry period at a hot, desert site (n ¼ 7 birds), and during a dry period (n ¼ 4 birds) and a wet period (n ¼ 5 birds) at a milder, semi-desert site. The desert birds maintained a significantly higher set-point T b (41.58 6 0.28C, mean 6 SD) than semi-desert birds (40.28 6 0.28C). During the warmest part of day (12:00-18:00 hours), T b increased significantly during periods of high T air and/or high humidity, and mean and maximum T b were up to 1.48 and 2.38C, respectively, above normal levels. However, as T air increased, birds at the desert site maintained T b at or below set-point levels for a greater proportion of the time than birds at the semi-desert site. Birds at the desert site also expressed a greater magnitude of daily heterothermy (heterothermy index, HI ¼ 2.48 6 0.38C, mean 6 SD) than birds at the semi-desert site: the latter population showed a greater magnitude of heterothermy during a dry period (HI ¼ 2.18 6 0.38C) than during a wet period (HI ¼ 1.68 6 0.28C). Birds continued foraging throughout the warmest part of the day, despite the fact that heat dissipation (percentage of time spent panting and wing-spreading) increased significantly with increasing T air . Our findings reveal that populations can vary in their thermoregulatory responses in both space and time and suggest that small changes in T air can have significant effects on thermoregulation in free-ranging desert birds, even when T air , T b . These data have important implications for assessing vulnerability of species to climate change, suggesting that sensitivity should be assessed at the population, rather than species, level.