Context. - Exercise hypoglycaemia (EH) is a very common yet little studied condition. It occurs in subjects with high values of insulin sensitivity (Se) and glucose effectiveness (S-G), and is mainly triggered by dietary errors, over-training, or special circumstances (exposure to unusual temperatures, altitude, disruption of circadian rhythms). Methods and results. - We showed that S-I and S-G can be calculated from test meals by mathematical modelling (oral minimal model), whereas in athletes "surrogates" of S-I (HOMA, QUICKI, etc.) are misleading (not applicable in this area of high insulin sensitivity). These meals can also provide a comprehensive picture of the components of insulin secretion, which makes this exploration more informative. We studied the results obtained by this method in 19 athletes complaining of EH compared to 58 matched athletes not complaining from this problem. We observed higher values of S-I (P < 0.05) and S-G (P < 0.03), but overall, insulin secretion Phi(oral) was similar. However, one of the components of insulin secretion, beta-cell glucose sensitivity, is higher on the average (P < 0.04) because in 30% of cases it remains above the normal range for athletes, while remaining in the range for a normal population. Therefore, overall tolerance to carbohydrates (product (Phi(oral) x S-I) is higher (+58% P < 0.04). Conclusion. - Thus, these subjects have a powerful carbohydrate disposal, which is incompletely compensated by the expected homeostatic decrease in insulin secretion, especially with regard to beta-cell glucose sensitivity. (C) 2011 Elsevier Masson SAS. All rights reserved.