Objective: The pharmacokinetics of a new insulin analogue, insulin aspart, were compared with unmodified human insulin in a double-blind crossover study of 25 fasting healthy men following a single subcutaneous dose. Methods: Either insulin aspart or human insulin, 0.1 U . kg-body-weight(-1), was injected subcutaneously and followed by determination of 8-h profiles of serum insulin and plasma glucose concentrations. Results: The absorption of insulin aspart was, on average, more than twice as fast and reached levels more than twice as high compared with human insulin [t(max(ins)) of 52 (23) vs 145 (93) min, P < 0.0001; and C-max(ins) of 41 (11) vs 18 (4) mU . 1(-1), P < 0.0001; mean with (SD)]. However, total bioavailability did not differ between the insulins, and thus the mean residence time was significantly shorter for insulin aspart [MRT(ins) of 149 (26) vs 217 (30) min, P < 0.0001]. Plasma glucose (PC) fell more than twice as rapidly [t(min(PG)) of 94 (45) vs 226 (120) min, P < 0.0001], to a greater extent [C-min(PG) 2.1 (0.6) VS 1.4 (0.4) mmol . l(-1), P < 0.0001], and for a shorter duration with insulin aspart than with human insulin. Conclusion: With improved subcutaneous absorption characteristics, the insulin aspart concentration-time profile resembles physiological meal-stimulated insulin release more closely than that of unmodified human insulin. This significantly alters the pharmacodynamic response in an advantageous manner in the meal-related treatment of diabetes mellitus.