We use a global chemical transport model (CTM) with size-resolved aerosol microphysics to evaluate our understanding of the processes that control Arctic aerosol, focussing on the seasonal changes in the particle size distribution during the transition from Arctic haze in spring to cleaner conditions in summer. This period presents several challenges for a global model simulation because of changes in meteorology, which affect transport pathways and precipitation scavenging rates, changes in the ocean-atmosphere flux of trace gases and particulates associated with sea ice break-up and increased biological activity, and changes in photolysis and oxidation rates which can affect particle nucleation and growth rates. Observations show that these changes result in a transition from an accumulation mode-dominated aerosol in spring to one dominated by Aitken and nucleation mode particles in summer. We find that remote Arctic aerosol size distribution is very sensitive to the model treatment of wet removal. In order to simulate the high accumulation mode concentrations typical of winter and spring it was necessary to substantially reduce the scavenging of these particles during transport. The resulting increases in accumulation mode lead to improvement in the modeled Aitken mode particle concentrations (which fall, due to increased scavenging in the free troposphere) and produce aerosol optical depths in good agreement with observations. The summertime increase in nucleation and Aitken mode particles is consistent with changes in local aerosol nucleation rates driven mainly by increased photochemical production of sulphuric acid vapor and, to a lesser extent, by decreases in the condensation sink as Arctic haze decreases. Alternatively, to explain the observed summertime Aitken mode particle concentrations in terms of ultrafine sea spray particles requires a sea-air flux a factor 5-25 greater than predicted by current wind speed and sea surface temperature dependent flux parameterizations. The enhanced total flux is clearly higher than measured in the Arctic and cannot explain the observed nucleation mode in the high Arctic. The model suggests that the summertime source of Aitken particles has very little effect on the accumulation mode and aerosol optical depth but they may contribute to cloud condensation nuclei in clouds with updraught velocities greater than about 15 cm/s. From a global aerosol modeling perspective, our understanding of Arctic aerosol is poor. We suggest several processes that currently limit our ability to simulate this challenging environment.