In this paper we employ renormalized viscosity and thermal diffusivity to construct a subgrid-scale model for large eddy simulation (LES) of turbulent thermal convection. For LES, we add to the kinematic viscosity; here is the turbulent kinetic energy flux, and is the grid spacing. We take subgrid thermal diffusivity to be same as the subgrid kinematic viscosity. We performed LES of turbulent thermal convection on a grid and compare the results with those obtained from direct numerical simulation (DNS) on a grid. We started the DNS with random initial condition and forked a LES simulation using the large wave number modes of DNS initial condition. Though the Nusselt number is overestimated in LES as compared to that in DNS, there is a good agreement between the LES and DNS results on the evolution of kinetic energy and entropy, spectra and fluxes of velocity and temperature fields, and the isosurfaces of temperature.