Capillary number (Ca), defined as dimensionless ratio of viscous force to capillary force, is one of the most important parameters in enhanced oil recovery (EOR). The ratio of viscous and capillary force is scale-dependent. At least 33 different Cas have been proposed, indicating inconsistencies between various applications and publications. The most concise definition containing velocity, interfacial tension and viscosity is most widely used in EOR. Many chemical EOR applications are thus based on the correlation between residual oil saturation (ROS) and Ca, which is also known as capillary desaturation curve (CDC). Various CDCs lead to a basic conclusion of using surfactant to reduce interfacial to ultra-low to get a minimum ROS and maximum displacement efficiency. However, after a deep analysis of Ca and recent new experimental observations, the traditional definition of Ca was found to have many limitations and based on misunderstandings. First, the basic object in EOR is a capillary-trapped oil ganglia thus Darcy's law is only valid under certain conditions. Further, many recent tests reported results contradicting previous ones. It seems most Cas cannot account for mixed-wet CDC. The influence of wettability on two-phase flow is important but not reflected in the definition of the Ca. Then, it is certainly very peculiar that, when the viscous and capillary forces acting on a blob are equal, the current most widely used classic Ca is equal to 2.2* 10−3. Ideally, the condition Ca ∼ 1 marks the transition from capillary dominated to viscous-dominated flow, but most Cas cannot fulfill this expectation. These problems are caused by scale dependent flow characterization. It has been proved that the traditional Ca is of microscopic nature. Based on the dynamic characterization of the change of capillary force and viscous force in macroscopic scale, a macroscopic Ca can well explain these complex results. The requirement of ultra-low IFT from microscopic Ca for surfactant flood is not supported by macroscopic Ca. The effect of increasing water viscosity to EOR is much higher than reducing IFT. Realizing the microscopic nature of the traditional Ca and using CDCs based on the more reasonable macroscopic Ca helps to update screening criteria for chemical flooding.