Using the perpendicular injection technique lean diabetic patients may often inject insulin intramuscularly (IM). Guided by ultrasound measurements of the subcutaneous (SC) thickness of the thigh, the aim of the present study was to re-evaluate the absorption kinetics of unmodified insulin from IM and SC injection sites and to evaluate the consequences of IM injection of unmodified insulin for blood glucose control in Type 1 diabetic patients. T50% values (time until 50% of the injected insulin is absorbed from the injection site) of SC injected, radioactively labelled, human unmodified insulin (125I-Actrapid) were 338 +/- 13 (+/- SE) min, 289 +/- 27 min, and 287 +/- 27 min during rest, light physical activity, and strenuous exercise, respectively. Intramuscularly injected unmodified insulin was absorbed faster, T50% 232 +/- 20 min, 113 +/- 13 min, and 112 +/- 5 min during the same levels of physical activity in the same order. When unmodified insulin (Actrapid) was given IM 30 min before breakfast, lunch, and dinner together with intermediate-acting insulin (Protaphane) SC at 2200 h, a more physiological profile of plasma free insulin and a more stable blood glucose profile was obtained than with SC administration into the thigh. The coefficient of variation of blood glucose concentration during the study (3 days each route) was lower with IM than with SC injection of unmodified insulin (33 +/- 4 vs 43 +/- 3%, p less than 0.01). No difference in frequency of hypoglycaemic attacks was found and patients claimed that IM injection was no more painful than SC injection. These data suggest that IM injection of soluble insulin into the thigh is beneficial.