Abstract Turbulent thermal mixing of fluids at different temperature in T-junctions represents a major concern for the safety of nuclear reactors. This is due to the significant thermal fluctuations that can arise under such circumstances, consequently leading to cyclic thermal stress and thermal fatigue within the pipe wall. Computational Fluid Dynamics (CFD) can be employed in order to obtain useful insights on the characteristics of the transient behaviour of the turbulent heat transfer in T-junctions. Although LES has been found to be the most accurate approach to turbulence modelling for this type of flow, its application at high Reynolds numbers is limited by its considerable computational costs. In this respect, the Unsteady Reynolds-Averaged Navier-Stokes (URANS) approach can be considered as a less computationally demanding option. In the present work, different URANS models are applied for the simulation of the thermal mixing in a T-junction at high Reynolds numbers. The numerical results are thoroughly assessed against the available experimental data. It is shown that, despite its limitations, a proper use of the URANS approach can give reasonable results for the considered flow configuration; in particular, a good prediction of the temperature fluctuations near the wall has been obtained, which is important for the evaluation of the cyclic thermal stress induced within the pipe wall. Therefore, it is concluded that URANS models can be regarded a pragmatic approach for the evaluation of temperature fluctuations in T-junction pipes. Finally, general guidelines for the application of the URANS approach for the simulation of such configurations are given.