The performance of three different configurations of a recently developed high volume cascade impactor (HVCI) system was tested in both field and laboratory experiments. In the field, the HVCI was run simultaneously with a virtual impactor and a low pressure impactor using a 3- or 4-day sampling duration. The study was carried out in downtown Helsinki between June and September 2001. As the field test results were not consistent with the previous laboratory calibration results provided by the manufacturer, the collection efficiency of each field-tested HVCI stage and two additional stages was measured in the laboratory as a function of particle size, using monodisperse dioctyl sebacate test aerosol. The collection efficiency curve for stage 1, impacting particles on a greased aluminum plate, was steep. In stages 2-4, polyurethane foam was used as collection substrate, which made the collection efficiency curves less steep than that of stage 1. According to the present laboratory calibration and field experiments, the HVCI configuration that suits best for physicochemical and toxicological characterization of ambient air aerosol particles consists of an Andersen high volume PM10-inlet and of three HVCI stages with the 50% collection efficiency at 2.4, 0.9 and 0.2 mum in aerodynamic diameter, respectively. (C) 2003 Elsevier Science Ltd. All rights reserved.