Plasma concentrations of glucose and insulin following a meal were compared in twelve Thoroughbred mares fed a pelleted concentrate (PC), a traditional sweet feed high in sugar and starch (SS), or a feed high in fat and fiber (FF). The feeds had similar DE and CP but differed in fat (19, 32, and 166 g/kg DM, respectively), NDF (199, 185, and 369 g/kg DM, respectively) and nonstructural carbohydrates (574, 645, and 247 g/kg, respectively). Mares were randomly assigned to two groups balanced for foaling date and weight. All mares received PC in late gestation; then, after foaling, one group was fed SS and the other FF for trials in early and late lactation. Mares were placed in stalls and deprived of feed overnight. A series of blood samples was collected via a jugular catheter from 0 (baseline) to 390 min after consumption of 1.82 kg of feed. Plasma was analyzed for glucose and insulin. Baseline values, peak values, and areas under curves (AUC) were compared by ANOVA. Baseline values were 74.7 +/- 10.9 mg/dL for glucose and 5.86 +/- 1.80 mIU/L for insulin for all diets and stages. Responses to PC did not differ between the two groups (P > 0.34), indicating the groups were metabolically similar. Peak plasma glucose and insulin concentrations were higher (P < 0.001) in the SS group than in the FF group during early and late lactation. Similarly, glucose and insulin AUC were larger (P < 0.003) in SS than in FF during early and late lactation. These results indicate that metabolic fluctuations are moderated by the replacement of sugar and starch with fat and fiber. This replacement may reduce the risk of certain digestive and metabolic disorders that have been linked to feeding meals of grain-based concentrates to pregnant or lactating mares.