Abstract. This paper presents an investigation of the robustness of correlations between characteristics of Arctic summer cyclones and September Arctic sea ice extent. A cyclone identification and tracking algorithm is run for output from 100-year coupled climate model simulations at two resolutions and for 30 years of reanalysis data, using two different tracking variables (mean sea-level pressure, MSLP; and 850 hPa vorticity) for identification of the cyclones. The influence of the tracking variable, the spatial resolution of the model, and spatial and temporal sampling on the correlations is then explored. We conclude that the correlations obtained depend on all of these factors and that care should be taken when interpreting the results of such analyses. Previous studies of this type have used around 30 years of reanalysis and observational data, analysed with a single tracking variable. Our results therefore cast some doubt on the conclusions drawn in those studies.