In order to remove xylene vapors from an air stream, an upflow laboratory scale biofilter was operated for a period of 2 months. The experimental study consisted of two different phases: in the first phase, the biofilter was operated at various gas flow rates and the xylene inlet concentration was maintained at 1.39 g m(-3). In the second phase, various inlet concentrations of the contaminant were tested at a constant gas flow rate of 0.4 m(3) h(-1) corresponding to an empty bed residence time of 150 s. The biofilter response to steep and abrupt variations in the xylene inlet concentration and gas flow rate was examined. The results obtained revealed that the removal efficiency of the biofilter regained its high values (above 96%) in less than 24h following the change to low concentrations and gas flow rate. Temperature measurements showed that the biofilter temperature strongly depends on the intensity of the microbial activity in the filter bed. The experimental mass ratio of carbon dioxide produced to the xylene removed was equal to 2.72 indicating that the contaminant was eliminated exclusively by aerobic biodegradation. These findings suggest that a follow up of the amount of carbon dioxide produced in the filter bed can be very helpful in monitoring the performance of the biofilter. For relatively small inlet loads of xylene, the contributions of the different sections of the biofilter to the removal efficiency of the contaminant and the carbon dioxide production were unevenly balanced but became more uniformly distributed for relatively high inlet loads. (C) 2004 Elsevier B.V. All rights reserved.