Abstract The aim of this work is to examine the capacity and the accuracy of a CFD (Computational Fluid Dynamics) model to characterize the thermo-aeraulic behavior of a heated room. Firstly, we present a brief description of the experimental set-up taken into account (two experimental tests with changed room boundary conditions were taken into account). Afterwards, the focus is on the main features of the numerical model (that strongly influence the accuracy of results): computational domain geometry and discretization, turbulence model, near wall treatment, radiation model and thermal boundary conditions. In addition, a simplified approach is presented here in order to integrate a pure buoyancy source within the model, based on a volumetric heat generation rate which is uniformly distributed within the heater. Furthermore, detailed experimental–numerical comparisons are given with regard to heat transfer to the walls as well as to heat source behavior and plume characteristics. The results obtained demonstrate that the CFD method employed in this work leads to reliable results. Consequently, this approach can be useful in detailed studies dealing with thermal comfort, indoor air quality and energy consumption for heated rooms. Finally, the simplified method presented here, concerning the integration of the heat source in the CFD model, can be effortlessly extended for other localized heat sources that differ in power, heat emission mode (convection or radiation), shape or size.