The hyperinsulinaemic-euglycaemic glucose clamp has always been regarded as the "gold standard" for the assessment of pharmacodynamic (PD) properties of insulin preparations; however, there has been controversy over a variety of methodogical details, such as study population, dosing time and the initial stabilization of blood glucose (BG) concentrations at the clamp target level, among clamp groups. As the impact of these details on PD results is unclear, the present review provides an overview of different methodological approaches for both the manual and the automated hyperinsulinaemic-euglycaemic glucose clamp. The advantages and limitations of several methodological details are discussed as well as the relevance of clamp results for the prediction of clinical outcomes. Overall, the best method strongly depends on the exact objective of the trial. If, for instance, duration of action is the primary objective, studies should be carried out in patients with type 1 diabetes to avoid any interference of endogenous insulin. This is less important for variables such as onset of action or early metabolic activity. The hyperinsulinaemic-euglycaemic glucose clamp has a high sensitivity to detect even minor differences between different insulin preparations. The practical relevance of potential differences, however, needs to be investigated in clinical studies. A major prerequisite for obtaining reliable glucose clamp results is the attainment of high clamp quality (i.e. keeping BG concentrations close to the clamp target throughout the experiments). Unfortunately, measures of clamp quality are often under-reported, as is the variability in PD profiles, although these might explain some unconfirmed extreme results obtained in a few clamp studies.