Turbulent Prandtl number (Pr) is a key parameter for controlling the temperature distribution in the research of thermal discharge water and other environmental hydraulics involving heat transfer. For a given problem, turbulent  Pr  number generally comes from previous experiences or trial-and-error method, which is blindfold and inefficient. To increase the reliability of turbulent  Pr  number for a given problem, Metropolis-Hastings algorithm, a Markov Chain Monte Carlo (MCMC) random sampling method was employed in this paper to identify turbulent Pr number. In the numerical simulation, steady standard k- ε model was used for turbulence flow field computation, while unsteady heat transfer equation was adopted for computing the temperature field. The computation results manifested that MCMC method is suitable and can offer precise results for the identification of turbulent Pr number.