Personalized ventilation (PV) aims to provide clean air to the breathing zone of occupants. Its performance depends to a large extent on the supply air terminal device (ATD). Five different ATDs were developed, tested and compared. A typical office workplace consisting of a desk with mounted ATDs was simulated in a climate chamber. A breathing thermal manikin was used to simulate a human being. Experiments at room air temperatures of 26 and 20 degreesC and personalized air temperatures of 20 degreesC supplied from the ATDs were performed. The flow rate of personalized air was changed from less than 5 up to 23 l/s. Tracer gas was used to identify the amount of personalized air inhaled by the manikin as well as the amount of exhaled air re-inhaled. The heat loss from the body segments of the thermal manikin was. measured and used to calculate the equivalent temperature for the whole body as well as segments of the body. An index, personal exposure effectiveness, was used to assess the performance of ATDs in regard to quality of the air inhaled by the manikin. The personal exposure effectiveness increased with the increase of the airflow rate from the ATD to a constant maximum value. A further increase of the airflow rate had no impact on the personal exposure effectiveness. Under both isothermal and non-isothermal conditions the highest personal exposure effectiveness of 0.6 was achieved by a vertical desk grill followed by an ATD designed as a movable panel. The ATDs tested performed differently in regard to the inhaled air temperature used as another air quality indicator, as well as in regard to the equivalent temperature. The results suggest that PV may decrease significantly the number of occupants dissatisfied with the air quality. However, an ATD that will ensure more efficient distribution and less mixing of the personalized air with the polluted room air needs to be developed.