The effects of task-ambient (TA) air-conditioning systems on the air-conditioning loads in a subway station and the thermal comfort of passengers were studied using computational fluid dynamics (CFD) and pedestrian-behavior simulations. The pedestrian-behavior model was applied to a standard subway station. Task areas were set up to match with crowdedness as predicted by the pedestrian-behavior simulations. Subsequently, a variety of TA air-conditioning systems were designed to selectively control the microclimate of the task areas. Their effects on the thermal environment in the station in winter were predicted by CFD. The results were compared with those of a conventional air-conditioning system and evaluated in relation to the thermal comfort of subway users and the air-conditioning loads. The comparison showed that TA air-conditioning systems improved thermal comfort and decreased air-conditioning loads. (c) 2005 Elsevier Ltd. All rights reserved.