Dairies in the United States commonly cool cattle with sprinklers mounted over the feed bunk that intermittently spray the cows' backs. These systems use potable water-an increasingly scarce resource--but there is little experimental evidence about how much is needed to cool cows or about droplet size, which is thought to affect hair coat penetration. Our objectives were to determine how sprinkler flow rate and droplet size affect physiological measures of heat load in a hot, dry climate, and to evaluate cooling effectiveness against water use. The treatments were an unsprayed control and 6 soaker nozzles that delivered four 3-min spray applications of 0.4, 1.3, or ≥ 4.5 L/min (with 2 droplet sizes within each flow rate) and resulting in 30 to 47% of spray directly wetting each cow. Data were collected from high-producing lactating Holsteins (n = 19) tested individually in ambient conditions (air temperature = 31.2 ± 3.8°C, mean ± standard deviation). Cows were restrained in headlocks for 1h and received 1 treatment/d for 3d each, with order of exposure balanced in a crossover design. When cows were not sprayed, physiological measures of heat load increased during the 1-h treatment. All measures responded rapidly to spray: skin temperature decreased during the first water application, and respiration rate and body temperature did so before the second. Droplet size had no effect on cooling, but flow rate affected several measures. At the end of 1h, 0.4 L/min resulted in lower respiration rate and skin temperature on directly sprayed body parts relative to the control but not baseline values, and body temperature increased to 0.2°C above baseline. When 1.3 or ≥ 4.5 L/min was applied, respiration rate was lower than the control and decreased relative to baseline, and body temperature stayed below baseline for at least 30 min after treatment ended. The treatment that best balanced cooling effectiveness against water usage was 1.3 L/min: although ≥ 4.5 L/min reduced respiration rate relative to baseline by 4 more breaths/min than 1.3 L/min did (-13 vs. -9 breaths/min, respectively), each additional liter of water decreased this measure by only ≤ 0.1 breaths/min (≤ 1% of the total reduction achieved using 1.3 L/min). We found similar water efficiency patterns for skin temperature and the amount of time that body temperature remained below baseline after treatment ended. Thus, when using this intermittent spray schedule in a hot, dry climate, applying at least 1.3 L/min improved cooling, but above this, additional physiological benefits were relatively minor.