Eight normal subjects were studied twice for 360 min after the subcutaneous injection of unmodified insulin (0.15 U kg-1) during euglycaemic clamps. Insulin absorption was assessed by both the area under the insulin-time curve above baseline (AUC) and time course of absorption (time to 25% and 50% of total AUC). Insulin action was measured as the amount of glucose infused. The maximal serum insulin concentration was 0.27 +/- 0.02 (+/- SE) nmol l-1 at 112 +/- 10 min. Fifty percent of total glucose infused occurred at 218 +/- 7 min. The maximal glucose infusion rate was 5.11 +/- 0.70 mg kg-1 min-1 and occurred at 256 +/- 12 min. Intrasubject coefficients of variation (CV) for total insulin AUC (11.2%), time to 25% of maximum AUC (12.1%) and time to 50% of maximum AUC (10.2%) were considerably lower than that for total insulin action (22.6%). Total insulin AUC did not correlate with total glucose utilization (r = 0.06, NS). We conclude that when glucose concentrations are maintained by euglycaemic clamps the peak of unmodified insulin action is later and the duration longer than traditionally recognized, insulin AUC does not predict insulin action, and the higher variability of insulin action compared with the indices of absorption suggests that day-to-day changes in tissue insulin sensitivity contribute more to the variability in insulin action than changes in absorption.