Although a hyperbolic relationship between insulin secretion and insulin sensitivity has been shown, the relationship has been often questioned. We examined the relationship using oral glucose tolerance test (OGTT)-derived indexes. A total of 374 Japanese subjects who had never been given a diagnosis of diabetes underwent a 75-g OGTT. In subjects with normal glucose tolerance (NGT), the ln [insulinogenic index (IGI)] was described by a linear function of ln (x) (x, insulin sensitivity index) in regression analysis when the reciprocal of the insulin resistance index in homeostasis model assessment, Matsuda's index, and oral glucose insulin sensitivity index were used as x. Because the 95% confidence interval of the slope of the regression line did not necessarily include -1, the relationships between IGI and x were not always hyperbolic, but power functions IGI x x(alpha) = a constant. We thought that IGI x x(alpha) was an appropriate beta-cell function estimate adjusted by insulin sensitivity and referred to it as beta-cell function index (BI). When Matsuda's index was employed as x, the BI values were decreased in subjects without NGT. Log BI had a better correlation with fasting plasma glucose (PG; FPG) and 2-h PG in non-NGT subjects than in NGT subjects. In subjects with any glucose tolerance, log BI was linearly correlated with 1-h PG and glucose spike (the difference between maximum PG and FPG). In conclusion, the relationship between insulin secretion and insulin sensitivity was not always hyperbolic. The BI is a useful tool in the estimation of beta-cell function with a mathematical basis.