This paper investigates the performance of a fuzzy reasoning machine for the control of indoor air quality in naturally ventilated buildings. Simulations have been performed using a new airflow and pollutant transport model, which has been developed and validated for this purpose; CO2 concentration was used as the indoor-air quality (IAQ) index for these simulations. Results have shown that satisfactory IAQ levels can be maintained, while good stability of the control parameter (i.e. window opening area) was achieved. The impact of such a controller on indoor-air temperature was also studied. The performances were not as good as expected, but were not negligible when compared with the normal conditions of use of the building.