This paper develops a stochastic model to assess storm risk in Austria, which relates wind speed and actual losses. By virtue of a building-stock-value-weighted wind index, we use suitably normalised historical loss data of residential buildings over 12 years and corresponding wind speed data to calibrate the model. Subsequently, additional wind speed data is used to generate further scenarios and to obtain loss curves for storm risk that give rise to storm insurance loss quantiles and corresponding solvency capital requirements both on the aggregate and on the regional level. We also investigate the diversification effect across regions and use tools from extreme value theory to assess the insurability of storm risk in Austria in general.