For the previous experiments of entrainment at T-junction, the air–cold water two-phase flow is always selected as working medium, and the theoretical models of liquid entrainment are established mainly based on the experimental data of air–water entrainment through a small break. The saturated steam–water entrainment experiments are gravely insufficient, especially the steam–water experiments of entrainment through a large size branch are even fewer. Hence, a scale-down experiment of steam–water entrainment at T-junction has been conducted referring to liquid entrainment through ADS-4 (Fourth Stage Automatic Depressurization System) in AP1000 nuclear plant. The steam–water experimental results are comprehensively compared with the results of the previous similar air–water experiment: The visualization research shows that the same entrainment phenomena can be observed including intermittent flow entrainment, stratified flow entrainment and annular flow entrainment; The transition conditions between the two entrainment phenomena have a little difference but can be negligible; The comparison of experimental data of entrainment onset and stable entrainment indicates that the experimental data of steam–water and air–water have the same change tendency, and the air–water correlation can agree well with the steam–water data. Hence, we can make a conclusion that effect of different working mediums on liquid entrainment is negligible. In addition, based on the analysis of model formation, we suppose that effect of the system pressure on entrainment can be neglected under 0.22MPa.