Five experiments were performed to test the overall hypothesis that exercise might be a useful indicator of growth hormone (GH) and prolactin status in horses. In experiment 1, geldings were exercised for 5 minutes four times at hourly intervals. The prolactin response (P < .05) to the first two exercise bouts was small and increased with successive bouts. There was a consistent GH response (P < .05) for only the first two bouts. In experiment 2, geldings were exercised for 29 to 39 minutes on a treadmill. After the initial bout, half the geldings were supplemented daily with Ca-β-hydroxy-β-methyl butyrate, and all geldings were conditioned for 12 weeks. Exercise bouts at 7 and 12 weeks indicated no effect (P > .1) of supplementation. In experiment 3, treatment of geldings with arginine before exercise increased (P < .001) prolactin concentrations but had no effect (P > .1) on the GH response to exercise. In experiment 4, the repeatability of the GH response to 5 minutes of exercise was determined by exercising eight stallions on six separate occasions. In addition to a large variation in GH response among stallions, there was a large variation within each stallion. In experiment 5, pretreatment with thyrotropin-releasing hormone 2 hours before exercise did not normalize the GH response to exercise. In conclusion, factors affecting the GH response to exercise likely preclude its usefulness as an indicator of GH status in horses.