Many thermal-hydraulic computer codes employ a fuel rod heat transfer model to couple the fuel rod temperatures with the hydraulic driving forces. Frequently, these models utilize uniform thermal conductivity for the fuel to reduce computer usage and storage. To evaluate the effect of this modeling, the uniform thermal conductivity model in COBRA III was modified to incorporate temperature-dependent thermal conductivity utilizing the complete expansion of the gradient of the heat flux, including the term that represents the gradient of the thermal conductivity. Demonstrative calculations for two transients showed that the peak fuel temperatures are very dependent upon the nonuniformity of the thermal conductivity. However, the peak cladding temperatures are almost independent of modeling of the thermal conductivity of the fuel because the clad temperatures are determined by the clad properties and the total amount of heat being transferred from the fuel to the coolant. The heat transferred is proportional to the integral of the thermal conductivity, which is virtually independent of the specific dependence of the temperature dependence of the thermal conductivity. The intermediate approach that employs the correct thermal conductivity at each point in the calculation but ignores the term in the heat conduction equation that accounts for the variation in the thermal conductivity was shown to yield results that are very similar to the uniform thermal conductivity cases. It is concluded that a uniform thermal conductivity model is adequate for models that are intended for the analysis of transients where the limiting constraint is the peak cladding temperature, such as the loss-of-coolant accident. However, models that are intended for the analysis of transients where the peak fuel temperature is limiting should employ the temperature dependence of the thermal conductivity.