To examine the reproducibility of the hyperinsulinaemic–euglycaemic clamp technique at mid-physiological hyperinsulinaemia, seven healthy subjects {age 50 (25, 59) years [median (range)], body mass index 23.1 (20.8, 25.5) kg·m-2} were investigated with three 2 h hyperinsulinaemic (60 µmol·l-1)–euglycaemic (4.5 mmol·l-1) clamps performed 48 h and 14 days apart respectively. The third clamp was prolonged to 8 h in order to examine effects on glucose disposal during prolonged clamps. The glucose infusion rates (GIRs) during the three 2 h clamps were 7.41 (4.28, 10.96), 7.26 (5.38, 11.02) and 6.63 (4.42, 10.3) mg·kg-1·min-1, with a median intra-individual coefficient of variation of 5.8 (2.6, 22) %. During the 8 h clamp a highly variable gradual increase in GIR was observed, reaching a plateau between 4 and 5 h at 32 (5, 101) % above the GIR between 1 and 2 h (P < 0.05). This increase was correlated inversely with the GIR between 1 and 2 h (r = -0.82; P < 0.05), and directly with age (r = 0.86; P < 0.05). Carbohydrate oxidation measured by indirect calorimetry was stable during the repeated 2 h clamps and the 8 h clamp. Endogenous glucose production measured by infusion of [6,6-2H2]glucose was suppressed during the 8 h clamp. The 2 h hyperinsulinaemic–euglycaemic clamp is reproducible at a mid-physiological range of hyperinsulinaemia. If prolonged, it results in a delayed increase in non-oxidative glucose disposal, which is most pronounced in subjects with low insulin sensitivity. The findings underline the importance of selecting age-matched controls in studies of insulin resistance.