It has been suggested that deviations from arterial euglycaemia during hyperinsulinaemic clamp experiments that use venous plasma glucose measurements may alter the results of such tests (glucose infusion rate, C-peptide suppression, etc.) and that 'arterialized' venous blood ('heated-hand' technique) may be suitable to circumvent these problems. Therefore, nine normal male fasting volunteers (age 25 +/- 4 years, body mass index 23.5 +/- 2.3 kg m-2) were examined three times using an insulin infusion of 1 mU kg min-1 over 120 min. Glucose was infused to maintain a concentration of 4.7 mmol I-1 (85 mg dl-1) in venous (V), 'arterialized' venous ('heated-hand' technique; HH), or capillary (C) plasma. The 'heated-hand' technique caused a rise in (rectal) body temperature of 0.3 +/- 0.1 degree C (P < 0.0001). Whereas the glucose aim was reached to a similar degree in all experiments (P = 0.36), capillary glucose concentrations differed slightly, but significantly (higher in experiments with venous and 'arterialized' venous blood specimens; P = 0.034). There were no significant differences regarding steady-state insulin concentrations (P = 0.77), glucose infusion rates (V, 7.1 +/- 0.5; HH, 7.2 +/- 0.6; C, 6.4 +/- 0.5 mg kg-1 min-1; P = 0.98), C-peptide suppression (P = 0.78), reduction in glucagon (P = 0.27) and free fatty acids (P = 0.16), and all parameters of indirect calorimetry (non-protein RQ: P = 0.67; glucose and lipid oxidation: P = 0.72 and 0.46 respectively; and energy expenditure: P = 0.42). Therefore, hyperinsulinaemic clamp experiments performed using venous, 'arterialized' venous, or capillary euglycaemia appear to be almost equally useful for the determination of insulin sensitivity and C-peptide or glucagon suppression. The elevation in body temperature that accompanies use of the 'heated-hand' technique does not noticeably influence measured metabolic parameters.