A numerical investigation of natural convection heat transfer is carried out for a single, horizontal, spent-fuel assembly in an environment typical of spent-fuel transportation systems as well as some dry storage/disposal scenarios. The objective is to predict computationally the convective heat transfer trends for horizontal spent fuel and to compare the results to data taken in a supporting experimental effort. The predicted data consist of thermal and flow fields throughout the assembly for a wide range of Rayleigh number, as well as numerically obtained Nusselt-number data that are correlated as a function of Rayleigh number. Both laminar and turbulent approaches are examined for a Boussinesq fluid with Pr = 0.7. The data predict the existence of a conduction-dominated regime, a transition regime, and a convection regime. Compared with the laminar approach, a significant improvement in the predicted Nusselt number is obtained for large Rayleigh numbers when a turbulence model is employed. This lends additional support to the experimental evidence that a transition to turbulent flow occurs for Rayleigh numbers greater than 107. Overall, the numerically predicted heat transfer trends compare well with previously obtained experimental data, and the computed assembly Nusselt numbers generally reside within the range of experimental uncertainty. The predicted thermal and flow fields further provide a numerical flow visualization capability that enhances the understanding of natural convection in horizontal spent fuel and allows improved physical interpretation of the experimental data.