This paper describes the concept and the first-stage development of a computational thermal manikin. The computational thermal manikin will be finally used to accurately predict the thermal sensation of a human in a room, by computer simulation. Human's thermal sensation directly depends on the heat transfer characteristics between the human body and his surrounding environment. The thermal environment around a modeled manikin will be examined by numerical simulation of airflow, thermal radiation, heat conduction, moisture transport, solar radiation and so on in a coupled way. In this paper, flow and temperature fields around a modeled manikin have been investigated with CFD (Computational Fluid Dynamics). The convective heat transfer characteristics of the modeled manikin are predicted, for several types of flow field, I.e., stagnant flow, horizontal uniform flow, downward uniform flow, and upward uniform flow. The CFD simulations are conducted using a low-Reynolds-number k-ε turbulence model. In order to model the complicated shape of the human body, a generalized curvilinear coordinate system (Boundary Fitted Coordinates) is used. The obtained results agree well with previous experimental data.