The thermal plume around a standing human subject was studied experimentally by Craven and Settles (JFE, 2007). The portion of the plume above the person's head was turbulent, and the maximum value of the time-averaged vertical velocity was 0.24 m/s; it occurred at a distance equal to 0.43 m directly above the subject's head. Craven & Settles also performed a RANS simulation of the thermal plume; the maximum value of the time-averaged vertical velocity obtained from the simulation was 0.20 m/s. This paper will present the results of numerical simulations of the thermal plume created by a seated mannequin. The simulations were performed using a lattice Boltzmann method. Following the detailed measurements of Craven and Settles, the surface temperature of the mannequin was taken to be 5oC warmer than the initial temperature of the surrounding air. Simulations were performed for three different spatial grids to determine the effect of spatial resolution on the velocity and temperature fields. In each case, a uniform, cubic spatial grid was used for the simulation; the grid spaces were 0.0067, 0.01, and 0.02m, respectively. The results obtained from the finest grid were in reasonable agreement with the experimental study; the maximum time-averaged vertical velocity was 0.226 m/s and it occurred at a distance 0.53m directly above the mannequin's head. The simulations also reveal that the instantaneous vertical velocity is significantly larger than the time-averaged vertical velocity, and that it occurs over a large range of distances above the mannequin; values as large as 0.34 m/s were obtained at times at late as 10 minutes from the initial condition. A more detailed comparison between the simulations and the experimental results will be presented in the talk.