OBJECTIVE To compare the accuracy and precision of cardiac output (CO) measurements derived from 4 thermodilution protocols that used different injectate temperatures and volumes in healthy adult horses. METHODS 8 healthy adult horses. METHODS Horses were anesthetized and instrumented with Swan-Ganz catheters. The CO was derived from each of 4 thermodilution protocols (IV injection of physiologic saline [0.9% NaCl] solution chilled to < 5 °C at volumes of 1 mL/15 kg of body weight [protocol A; control], 1 mL/25 kg [protocol B], and 1 mL/35 kg [protocol C] or maintained at 17 °C at a volume of 1 mL/15 kg [protocol D]) 3 times during each of 5 measurement cycles, with a 30-minute interval between cycles. During each measurement cycle, protocol A was performed first, and protocols B, C, and D were performed in a randomized order. Mean CO and within-subject variance in CO were compared among the 4 protocols. RESULTS Mean CO did not differ significantly among the 4 protocols. The within-subject variance for CO measurements derived from protocols C and D, but not protocol B, was significantly greater than that for protocol A (control). CONCLUSIONS Results suggested that, in healthy adult horses, decreasing the thermodilution injectate volume to 1 mL/25 kg from the recommended volume of 1 mL/15 kg did not adversely affect the accuracy or precision of CO measurements. However, use of smaller injectate volumes or use of injectate at approximately room temperature is not recommended owing to a clinically unacceptable increase in CO measurement variability.