There is a need for reliable measurements of insulin sensitivity (SI) simpler than the euglycemic hyperinsulinemic clamp or the intravenous glucose tolerance test (IVGTT), which could be used when the simpler surrogates based on fasting insulin (Ib) and glucose (Gb) lose their validity. Several evaluations of SI derived from oral glucose tolerance test (OGTT) or its physiologic form, the standardized breakfast test (SBT), have been proposed. We aimed at determining which SBT-derived measurements of SI give the best prediction of the values obtained with the minimal model analysis of an IVGTT. Twenty-eight subjects (23 females and 5 males; age, 44.3+/-0.6 years) with a wide range of glucose tolerance randomly underwent a hyperglucidic SBT and an IVGTT with minimal model analysis. Correlations of 35 indices (converted if appropriated into similar units) with IVGTT-derived SI were calculated, and the accuracy of the empiric formulas obtained with the 11 best predictions were evaluated with Bland-Altman plots. Subjects covered all the spectrum of SI between 0.19 and 21.3 min-1/(microU.mL-1)x10(-4). Eight procedures yielded satisfactory predictions of minimal model SI: (1) SI (from Matsuda's composite index)=-1.24+65/(IbGbImGm)-0.5; (2) SI=1.89+2690/(IbGbImGm); (3) SI (from Bennett's index)=-2.93+5.16/(log Ibxlog Gb); (4) SI (from Sluiter's index)=0.2+2400/(IpGp); (5) SI=-8.54+38.4/(Belfiore's ISI index); (6) SI (from Cederholm's formula)=76/(Gm log Im); (7) SI=0.248+0.947/GbIm; (8) SI (from Mari's "oral glucose insulin sensitivity" index)=oral glucose insulin sensitivity/Ip; (9) Caumo's model. Glucose effectiveness Sg can also be accurately predicted by the following formula: Sg=2.921e-0.185(G60- Gb) (Ip=insulin peak; Gp=glucose peak; Ia=insulin area; Ga=glucose area; G60=glycemia at 60 minutes). The hyperglucidic SBT can provide accurate evaluations of SI and Sg, either by elaborated models or by simple empiric formulas.