In order to simulate actual behavior of bleed air flows in various air-intake operating conditions, it is necessary to consider the change in bleed mass flow rates according to the bleed plenum conditions. Wind tunnel testing was carried out for validation of the conventional Harloff’s model for the rates. The results showed that the model overestimates the bleed mass flow rates in certain cases where blowing from the bleed plenum into the main stream occurs. A new boundary condition model improved by addition of the effect of the main stream dynamic pressure is proposed and enables quantitative prediction of the bleed mass flow rates in such cases.