The model of interfacial heat transfer for the dispersed flow regime used in the RELAP5 computer codes is investigated in the present paper. Short-transient calculations of two low flooding rate tube reflooding experiments have been performed, where the hydraulic conditions and the heat input to the vapour in the post-dryout region were controlled for the predetermined position of the quench front. Both RELAP5/MOD2.5 and RELAP5/MOD3 substantially underpredicted the exit vapour temperature. The mass flow rate and quality, however, were correct and the heat input to the vapour was larger than the actual one. As the vapour superheat at the tube exit depends on the balance between the heat input from the wall and the heat exchange with the droplets, the discrepancy between the calculated and the measured exit vapour temperature suggested that the inability of both codes to predict the vapour superheat in the dispersed flow region is due to the overprediction of the interfacial heat transfer rate.