This paper discusses the implementation of a burnup dependent fuel thermal conductivity model within the Reactor Dynamics and Fuel Modeling Group (RDFMG) version of the subchannel thermal-hydraulics code COBRA-TF (CTF). The model takes into account the degradation of fuel thermal conductivity at high burnups and its dependence on the Gadolinium content for both UO2 (uranium dioxide) and MOX (mixed oxide) nuclear fuel rods. The modified Nuclear Fuel Industries (NFI) model for UO2 rods and the Duriez/Modified NFI model for MOX rods were incorporated in CTF. To validate the models, the fuel centerline temperatures predicted with CTF were compared to Halden reactor experimental data and to high fidelity FRAPCON-3.4 calculations. Halden test cases for UO2 fuel rods at the beginning of life (BOL), through lifetime with and without Gd2O3; and for MOX fuel rods were simulated with CTF. It was demonstrated that CTF with the new burnup dependent fuel thermal conductivity model predicts the fuel centerline temperature with less than a 5% error as compared to the Halden measurements. CTF calculations were performed for fifty-eight (58) data points. Statistical analyses of the dimensionless predicted-to-measured fuel centerline temperature ratios had confirmed the advantage of the new model – the mean value of the predicted-to-measured temperature ratios was increased from 0.8920 to 1.0082 and the standard deviation was decreased from 0.0693 to 0.0382.