Aims/hypothesis. We hypothesized that beta-cell responses to changes in glucose would not be normal in subjects with impaired glucose tolerance (IGT). Methods. Three groups of 6 subjects were studied: normal weight with normal glucose tolerance (control subjects); obese with normal glucose tolerance (Obese-NGT); and obese with IGT (Obese-IGT). All subjects had a graded glucose infusion protocol to increase (step-up) and then decrease (step-down) plasma glucose. We obtained average insulin-secretion rates (ISR) over the glucose range common to all three groups during step-up and step-down phases, minimal model indices of beta-cell function (f_b, f_d, f_s, T_up, T_down ), and insulin sensitivity (Si). Results. ISR differed significantly between step-up and -down phases only in Obese-IGT individuals. Basal (f_b) and stimulated (f_d, f_s) beta-cell sensitivity to glucose were similar in the three groups. Delays between glucose stimulus and beta-cell response during both step-up (T_up) and -down (T_down) phases were higher in Obese-IGT compared to Controls and Obese-NGT individuals. The product ISR × Si (10^–5·min^–2× l) was lower in Obese-IGT compared to Controls, both during step-up (919 ± 851 vs 3192 ± 1185, p < 0.05) and step-down (1455 ± 1203 vs 3625 ± 691, p < 0.05) phases. Consistently, the product f_s× Si (10^–14·min^–2· pmol^–1× l) was lower in Obese-IGT than in control subjects (27.6 ± 25.4 vs 103.1 ± 20.2, p < 0.05). Conclusion/interpretation. Subjects with IGT are not able to secrete insulin to compensate adequately for insulin resistance. They also show delays in the timing of the beta-cell response to glucose when glucose levels are either rising or falling. [Diabetologia (2002) 45: ▪–▪]