Abstract. Airborne measurements of aerosol particles and trace gases were performed at midlatitudes in the Stratosphere-Troposphere Experiment by Aircraft Measurements (STREAM-96) campaign during late spring in the free troposphere and lower stratosphere. The results show a consistent relation between particle number concentration and water vapor mixing ratio, suggesting a minimum particle number concentration below which nucleation of particles is very likely to occur. Numerical simulations, using a coupled atmospheric chemistry and aerosol nucleation model, support this hypothesis. The model calculations indicate that new particle formation is very sensitive to the preexisting particle number concentration in the tropopause region. To be able to quantify and predict future aerosol effects regarding climate, chemistry, and human health, it is essential to improve our understanding about how particles form, transform, and are removed from the atmosphere. Of these three processes, particle formation is perhaps the least understood.Numerous events of particle formation have been reported, most frequently from observations in the marine boundary The most important mechanism for new particle production ;o thought to be binary homogeneous nucleation of sulphuric acid and water vapor. At a certain critical supersaturation of sulphuric acid the vapor condenses and forms new particles. At increasing relative humidity this threshold is significantly reduced, and particle formation can occur at much lower H2SO 4 concentrations [Kulmala et al., 1995]. Normally, the concentration of n2so 4 vapor in the atmosphere is controlled by the balance between the source, that is, oxidation of SO2 by OH, and the sink, condensation on preexisting particles. The available aerosol surface area and number concentration determines the rate of condensation [Pirjola and Kulmala, 1998]. Thus a low preexisting particle number concentration and surface area, high humidity, and high precursor gas concentrations (OH and SO2) favor this type of particle formation. Observations by Brock et al. [1995] showed that particles in the upper tropical troposphere were composed of H2SO4-H20 droplets and thus were very likely formed by homogeneous nucleation of sulphuric acid and water vapor. Covert et al.[1992] reported events of new particle formation in the marine boundary layer connected to periods with low aerosol surface area at times when there was no SO2 enhancement. An inverse relationship between surface area and ultrafine particles was also observed by Clarke [1992] in the upper troposphere. Hegg et al. [1990] found new particle production just above the cloud top in the marine boundary layer, presumably due to increased relative humidity caused by the evaporation of cloud droplets. Cases of new particle formation in the marine boundary layer by oxidation of SO2 and subsequent condensation of H2SO 4 have been shown by Hoppel et al. [1994]. However, the sulphuric acid concentrations were probably well below the binary 31,255