Assessment of the allergen content of airborne particles small enough to reach the bronchiolar airways is important for a better understanding of allergic asthma. In order to test the performance of a high-volume cascade impactor for size-selective sampling of airborne particles, the characteristics of pollen deposition, and the particle size-dependent allergen distribution, were re-examined. Two cascade impactors with rectangular slots were run in parallel, one with glass fibre filters on all stages, the other with silicone grease on stage 2 (collection of particles of size 4.2-10.2 micrometer) instead. Pollen was counted using light microscopy and allergens were measured using ELISA techniques. In the impactor without the greased stage 2, a heavy bounce and blow-off for pollen was found. Bounced pollen was deposited mainly on the back-up filter, the sampling stage for particles smaller than 1.4 micrometer. However, if stage 2 was coated with silicone grease, less than 1% of total pollen was found in the lower stages. On the first stage of the sampler (50% cut-point, 10.2 micrometer) where all pollen should impact, pollen had a deposition efficiency of only 24-47%, depending on the abundant pollen species. Much less allergen was found in fine particle fractions than in previous studies in which particles were sampled with similar samplers but without grease as a pollen trap. The observed bounce and blow-off of pollen without a greased stage 2 leads to serious mistakes when assessing the allergen content of the fractions collecting particles smaller than intact pollen, i.e., below 10.2 micrometer. Pollen allergen concentrations in respirable particle fractions are much lower than in the pollen size fraction. This is of great importance both for symptoms of sensitized allergic asthmatics and for research of asthma-related mechanisms.