Flooding in the process of well killing in empty wells results in a waste of kill fluid and increases the time of well killing, which is not conducive to its smooth operation. To address this problem, a large-size vertical visualization experimental device was designed and constructed, falling experiments were performed on kill fluid under different apparent gas velocities, apparent liquid velocities, and kill-fluid viscosities, and the gas–liquid two-phase migration state at the injection point during the process of flooding was analyzed in this study. Commonly used critical gas velocity models of flooding were compared, considering the influence of fluid properties and rheological parameters on the critical gas velocity. A combination of dimensionless numbers was used to fit the experimental data, and a model of the critical gas velocity of flooding was established. The comparison results showed that the prediction accuracy of the relationship in this study was higher than that of the traditional relationship, with an average error of 9.89%. The influence of different parameters on the critical gas velocity of flooding was analyzed, and the key parameters for well killing in empty wells were optimized, providing a theoretical basis for empty well killing operations.