We report on computational fluid dynamics (CFD) predictions of mixing time of a pollutant in an unventilated, mechanically mixed, isothermal room. The study aims to determine: (1) the adequacy of the standard Reynolds Averaged Navier Stokes two-equation (k−ε) turbulence model for predicting the mixing time under these conditions and (2) the extent to which the mixing time depends on the room airflow, rather than the source location within the room. The CFD simulations modeled the 12 mixing time experiments performed by Drescher et al. (Indoor Air 5 (1995) 204) using a point pulse release in an isothermal, sealed room mechanically mixed with variable power blowers. Predictions of mixing time were found in good agreement with experimental measurements, over an order of magnitude variation in blower power. Additional CFD simulations were performed to investigate the relation between pollutant mixing time and source location. Seventeen source locations and five blower configurations were investigated. Results clearly show large dependence of the mixing time on the room airflow, with some dependence on source location. We further explore dependence of mixing time on the velocity and turbulence intensity at the source location. Implications for positioning air-toxic sensors in rooms are briefly discussed.