The choice of an arterial cannula for cardiac surgery is often difficult. The clinician has to rely on the pressure–flow diagrams provided by the manufacturer that are all produced using water. The conversion of these water data to representative clinical conditions is often difficult if not impossible. In this in vivo study the theory of dynamic similarity is applied to scale the manufacturer's data to the clinical situation, in order to predict the pressure drop desired by the cardiac surgeon. Three hundred and forty-seven samples of pressure drop are obtained in 58 patients. Patients are grouped according to cannula size (20, 22, and 24 Fr). Very good correlation is found between measured and predicted values (r = 0.93; 0.86; 0.87 for 20; 22; 24 Fr). As a result, windows of optimal performance are constructed for each cannula diameter. This technique also allows the construction of an online recommendation for signaling abnormal cannula performance during cardiac surgery.