In June 1988, during the grass-pollen season in Leiden, The Netherlands, outdoor airborne paniculate matter was collected and separated into fractions according to aerodynamic sizes ( 10μm, 4·9–10μm, 2·7–4·9μm, 1·3–2·7μm, 0·6–1·3μm, 0·6μm), with a cascade impactor mounted on top of a high volume sampler. The different fractions were tested for the presence of grass-pollen allergenic activity using a RAST-inhibition assay: specific IgE-antibody-containing patient serum was applied on the particle-loaded impaction strips, and the serum was recovered by descending elution for further analysis in the RAST. Simultaneously, continuous measurements were made of the airborne grass-pollen concentration using a volumetric pollen trap. Sampling observations lasting 7–9 hr during a period with relatively high airborne grass-pollen concentrations showed reliably detectable amounts of grass-pollen allergen, not only in the first impaction stage where intact pollen were collected, but also in the lower stages collecting the smaller, paucimicronic and submicron atmospheric aerosol fraction. It is evident that this result has serious implications for the understanding of the bronchial symptoms frequently seen in hay fever patients on days with high pollen concentrations in the air.