Four experiments were conducted 1) to assess the use of glucose responses to insulin injections as a means of estimating insulin sensitivity in horses and 2) to compare the insulin sensitivities of normal horses vs. those displaying hyperleptinemia (HL). In Exp. 1, HL mares and geldings (n = 4 each) and 4 mares and geldings with normal leptin concentrations (NL) were injected intravenously with 20 and 100 mU/kg of BW of bovine insulin on 2 separate occasions in December 2008. In Exp. 2, the experimental protocol was repeated in late April 2009. In Exp. 1, the 20 mU/kg of BW dose of insulin caused a greater (P < 0.05) decline in glucose concentrations in NL mares and geldings compared with HL horses. The response of HL mares to the 100 mU/kg of BW dose was less (P < 0.05) than for the other groups. In Exp. 2, responses of all groups to the 20 mU/kg of BW dose were small and similar among groups (P > 0.1), whereas the greater dose revealed differences (P < 0.05) in sensitivity among groups consistent with those observed with the smaller dose in Exp. 1. Experiment 3 was conducted in June and July of 2009 to further examine the dose-response relationship in mares of potentially different insulin sensitivities in an attempt to standardize the approach for studying a wide range of sensitivities. Recombinant human insulin was used at doses of 8, 20, 50, and 125 mU/kg of BW, as needed, to estimate (by linear regression) the dose of insulin causing a 50% decline in glucose concentrations (ED50). Five mares each of reduced leptin concentrations (LL) and small BCS (3 to 5), LL and larger BCS (6 to 7.5), and increased leptin concentrations and increased BCS were studied. The ED50 was similar (P > 0.1) for LL mares, regardless of BCS, and was less (P < 0.01) than for mares with increased leptin concentrations. It was concluded that a dose of 50 mU/kg of BW of recombinant human insulin could be used safely to start the dose-response curve; smaller or larger doses could then be applied as appropriate to get sufficient data for estimation of ED50. Experiment 4, conducted in October of 2009, assessed the repeatability of the estimates for ED50 obtained in Exp. 3. Six mares with LL vs. increased leptin concentrations received the 50 mU/kg dose of insulin; appropriate larger or smaller doses were used to obtain estimates of ED50. Estimates obtained were highly correlated (r = 0.91) with those obtained in Exp. 3, with an average within-mare CV of 8.9%; this is equal to or better than the repetabilities of the currently used methods of assessing insulin sensitivity in horses. It was concluded that hyperleptinemic horses, which are also hyperinsulinemic and have exaggerated insulin responses to glucose injection, are indeed less sensitive to insulin than normal horses with reduced leptin concentrations of the same body condition.