Heat transfer coefficients have been computed for flow in a pipe and flow between two plates with correlations and turbulence models based on Reynolds Averaging of the Navier-Stokes (RANS) equations. Predictions of the correlations and those of RANS turbulence models have been compared with experimental data of flow in a pipe. The correlations considered are those of Dittus-Boelter, Seider-Tate, Petukhov, and Sleicher-Rouse, while the turbulence models include the standard high Reynolds number, the Reynolds stress model, the low Reynolds number, and the v2f model. There are significant differences in the predictions of the correlations as well as in those of the turbulence models. Although computational fluid dynamics simulations have wider applicability and provide more information than simulations using correlations, the heat transfer coefficient predicted by the turbulence models is not always more accurate than that predicted by correlations. The discrepancy in the heat transfer coefficient predicted by the turbulence models is due mainly to discrepancies in the prediction of turbulence near the wall and to the uncertainty in the value of the turbulent Prandtl number.