The aim of this study was to quantify the effect of the interaction between 2 constant ambient temperatures [thermoneutrality (TN; 15°C) and high temperature (HT; 28°C)] and 2 levels of Na bicarbonate supplementation [calculated to provide diet Na contents of 0.20%DM (Na−) and 0.50%DM (Na+)] on water partitioning in dairy cows. Treatments were compared on 4 dry and 4mid-lactation Holstein cows according to 2 Latin squares (1 for each physiological stage) over the course of 4 periods of 15d. Diets consisted of a total mixed ration based on maize silage. Dry cows were restricted to their protein and energy requirements, whereas lactating cows were fed ad libitum. The daily average temperature-humidity index was 59.4 for TN and 73.2 for HT. Lactating and dry cows had higher vaginal temperatures at HT than at TN, but the increase was more pronounced in lactating cows (+1.05 vs. +0.12°C for vaginal temperature, respectively). Dry matter intake (DMI) of lactating cows decreased by 2.3kg/d at HT. Free water intake (FWI) and estimated volume of water lost to evaporation increased at HT in both lactating and dry cows; no interactions were observed between temperature and physiological stage. When expressed as a proportion of DMI, the increase in evaporation that occurred with increasing temperature was completely compensated for by an increase in FWI for both physiological stages. The urinary water excretion increased slightly at HT in lactating cows but not in dry cows, which may be related to the low chloride content of the offered diet. High Na supplementation increased DMI slightly in lactating cows, but milk yield was not affected. Sodium supplementation did not limit the decrease in DMI observed in lactating cows at HT; this observation is likely due to the high diet electrolyte balance of the offered diets. Sodium supplementation increased FWI in lactating cows and urinary flow in both physiological states. The interaction between ambient temperature and Na supplementation did not affect either water intake or water evaporation. This study demonstrates that the development of predictive models for water intake that include environmental variables could be based on mechanistic models of evaporation.