The present state of knowledge unequivocally indicates that chronic diabetes is associated with impaired function of coronary vessels. Langendorff retrograde perfusion is one of the most frequently employed methods to study dysfunction of coronary vasculature in animal models of diabetes mellitus. However, because of methodological discrepancies in experimental protocols, the reliability of this technique is limited. In the current study, we propose the novel technique of vasoactive drug administration and aim to evaluate its usefulness in detecting coronary dysfunction in diabetes. Using Langendorff model, we compared the results of coronary endothelium-dependent (bradykinin) and -independent (diethylamine/nitric oxide, DEA/NO) vasodilatation obtained from experimental model utilizing automatically corrected-rate infusion with commonly used, constant-rate infusion of vasoactive drug. The infusion of bradykinin at constant rate failed to reveal coronary endothelium-dependent dysfunction typical for diabetes mellitus. Induction of endothelium-independent vasodilatation by constant infusion demonstrated augmented response in diabetic hearts. The administration of bradykinin or DEA/NO at the corrected rate was associated with significantly increased maximal responses in comparison with constant infusion experiments. This phenomenon was observed particularly in the control group. We conclude that only corrected-rate infusion of vasoactive agents to actual value of coronary flow enables the reliable detection of endothelial dysfunction in diabetes mellitus.