This study tested the hypothesis that exercise training would alter feed intake (FI), body composition (BC) and plasma concentrations of active ghrelin, leptin, cortisol, insulin and glucose. Eight Standardbred mares (12±2 years, 509±36 kg body weight (BW), mean±SD) were trained (EX) in an equine Equi-ciser (initially 3 days per week at 60% maximal heart rate (HRmax) for 20 min and gradually increased to 5 days per week at 70% HRmax for 30 min, with a 10-min warm-up and 10-min cool-down period at the walk). Six mares (12±2 years, 537±45 kg) served as non-exercise controls (CON). All mares were unfit and had not been subjected to conditioning for 3 years before the experiment. Pre- and post-training incremental exercise tests (GXT) were run to determine HRmax and maximal oxygen uptake (VO2max). A total mixed ration (TMR) of hay cubes was fed free choice for 16 h day−1 with the primary experiment following a 6-week diet adaptation period. Mares' FI was measured daily and reported in grams per kilogram BW of feed eaten per week. Changes in BC were assessed using BW (electronic scale) and percentage fat calculated using rump fat thickness and the Westervelt equation. Blood samples were taken every 2 weeks at 15:25, before mares were given their allotment of hay cubes on a day when they did not exercise, to measure plasma hormone and glucose concentrations. Gastroscopy for gastric ulcers was performed before, during and after the trial. VO2max increased by 7.0% (P<0.03) in EX, but did not change (P>0.05) in CON. FI decreased (P<0.001) in both groups, but was only different (P<0.02) between EX and CON at week 3. Digestible energy (DE) intake (Mcal day−1) was initially higher (P<0.001) than calculated DE requirements in EX. However, over time, DE only matched and then fell below (P<0.03) the DE intake required for training. In CON horses, DE intake was higher (P<0.001) than calculated requirements. BW and percentage body fat increased (P<0.001) over time in EX and CON. Plasma leptin concentration increased (P<0.001) over time in both groups, but was 60% higher (P<0.04) in CON compared to EX at weeks 4–8. There were no differences (P>0.05) in active ghrelin, glucose, insulin or cortisol between the groups and over time. Five out of seven EX mares developed gastric ulcers. No CON mares developed gastric ulcers. Training was associated with changes in plasma leptin concentration, an increased incidence of gastric ulcers and a disruption of the balance between required DE and actual intake.