The purpose of this work was to recommend a method of measuring the adsorptive capacity of powdered activated carbon (PAC). The adsorptive capacity is needed in future mathematical models of dispersed phase adsorption. This measurement was difficult, because the large drag forces in powders resist any flow into the sample volume, causing big pressure gradients. Two approaches were tested and compared; a material balance over a packed bed, and a standard volumetric method. The pressure loss across the packed bed was made small by the combination of the PAC with glass ballotini. Toluene was chosen as a sorbate because of its ease of measurement. The results concerned pressure variations in the packed bed, and the ease with which the end-points could be found in either method. With regard to the packed bed, the time from the start of breakthrough to saturation was typically 110 min. The principal reason for this delay was channelling; the breakthrough curve could not be used to infer pore diffusion coefficients. The total pressure within the bed was kept to within 5% of its mean value. Hence, the adsorptive capacity could be plotted as a function of pressure, and fitted with a Freundlich isotherm with an exponent of n = 0.2. Including the PAC, the adsorptive capacities of four kinds of carbon, at the same temperature and pressure, varied from 8 to 16%. For the volumetric method, at least 3 h were needed for the sorbent and sorbate to reach equilibrium. A mathematical model showed that the uptake curve was controlled by heat transfer. Because the test volume was under a partial vacuum, the technician had to be careful that no air leaked into it during the test. The volumetric method was faster with 250 mum diameter carbon granules; the end-point was evident after 20 min. In conclusion, a method involving a packed bed was developed to measure the adsorptive capacity of a powder. The end-point was determined easily; the end-point for the volumetric method was indistinct. (C) 2004 Elsevier B.V. All rights reserved.