To determine the effect of dietary starch, bicarbonate, and fat content on metabolic responses and serum creatine kinase (CK) activity in exercising Thoroughbreds with recurrent exertional rhabdomyolysis (RER), 5 RER horses were fed 3 isocaloric diets (28.8 Mcal/d [120.5 MJ/d]) for 3 weeks in a crossover design and exercised for 30 minutes on a treadmill 5 days/wk. On the last day of each diet, an incremental standardized exercise test (SET) was performed. The starch diet contained 40% digestible energy (DE) as starch and 5% as fat: the bicarbonate-starch diet was identical but was supplemented with sodium bicarbonate (4.2% of the pellet): and the fat diet provided 7% DE as starch and 20% as fat. Serum CK activity before the SET was similar among the diets. Serum CK activity (log transformed) after submaximal exercise differed dramatically among the diets and was greatest on the bicarbonate-starch diet (6.51 +/- 1.5) and lowest on the fat diet (5.71 +/- 0.6). Appreciable differences were observed in the severity of RER among individual horses. Postexercise plasma pH, bicarbonate concentration, and lactate concentration did not differ among the diets. Resting heart rates before the SET were markedly lower on the fat diet than on the starch diet. Muscle lactate and glycogen concentrations before and after the SET did not differ markedly among the diets. A high-fat, low-starch diet results in dramatically lower postexercise CK activity in severely affected RER horses than does a low-fat, high-starch diet without measurably altering muscle lactate and glycogen concentrations. Dietary bicarbonate supplementation at the concentration administered in this study did not prevent increased serum CK activity on a high-starch diet.