Reliable physiological markers for performance evaluation in sport horses are missing. To determine the diagnostic value of plasma ACTH and cortisol measurements in the warmblood horse, 10 initially 3-yr-old geldings of the Hannovarian breed were either exposed to a training schedule or served as controls. During experimental Phase 1, horses were group-housed, and half of the horses were trained for 20 wk on a high-speed treadmill. During Phase 2, groups were switched and one group was trained for 10 wk as during Phase 1, whereas the control group was confined to boxes. During Phase 3 horses were initially schooled for riding. Thereafter, all horses were regularly schooled for dressage and jumping, and half of the horses received an additional endurance training for 24 wk. During all phases horses were exposed at regular intervals to various standardized treadmill exercise tests. During and after the tests frequent blood samples were taken from an indwelling jugular catheter for determination of ACTH and cortisol. Treadmill exercise increased both hormones. Maximum ACTH concentrations were recorded at the end of exercise, and maximum cortisol levels were recorded 20 to 30 min later. Except for one test there were no differences in ACTH levels between trained horses and controls. There was no significant effect of training on the cortisol response (net increase) to treadmill exercise in any of the tests during Phase 1. During Phase 2 higher cortisol responses were recorded in controls than in trained horses (P < .05) after 10 wk of training (controls confined to boxes). During Phase 3 plasma cortisol responses were also higher in controls than in trained horses (P < .05 after 6, 18, and 24, P < or = .07 after 12 wk of training) when the inclination of the treadmill was 5%, but not at 3%. There was no overlap in net cortisol responses at 30 min between trained and untrained horses. An ACTH application after 24 wk of training resulted in higher cortisol responses in controls than in trained horses (P < or = .05), without any overlap between the groups at 30 min after ACTH. Plasma cortisol responses to either treadmill exercise or ACTH injection may be a reliable physiological marker for performance evaluation. Prerequisites are sufficient differences in training status and sufficient intensity of exercise test conditions.