Abstract: This study investigated three methods of water supply on drinking preference and behavior in six Standardbred geldings (2-9 years, 505+/-9 kg). The water sources were buckets (B), pressure valve (PV), and float valve (FV) bowls. In an initial drinking preference test, PV was tested at three flow rates: 3, 8, and 16 l/min (PV3, PV8, and PV16), and FV at 3 l/min (FV3). Water intake was measured in l and presented as the percentage of the total daily water intake from each of two simultaneously presented alternatives. The intake from PV8 was greater than from both PV3 (72+/-11% vs. 28+/-11%) and PV16 (90+/-4% vs. 10+/-4%). All horses showed a strong preference for B, 98+/-1% of the intake compared to 2+/-1% from PV8. Individual variation in the data gave no significant difference in preference between the two automatic bowls. In the second part of the study, drinking behavior and fluid balance were investigated when the horses drank from FV3, PV8, and B for 7 consecutive days in a changeover design. Despite a tendency for an increase in total daily drinking time from FV3, the daily water intake was significantly lower (43+/-3 ml/kg) than from PV8 (54+/-2 ml/kg) and B (58+/-3 ml/kg). Daily net water gain [intake-(fecal+urinary output)] was only 0.5+/-3 ml/kg with FV3, resulting in a negative fluid balance if insensible losses are included. These results show that the water supply method can affect both drinking behavior and fluid balance in the horse.