Catastrophe (‘cat’) loss modelling is used to assess the risk of natural hazards by leveraging, among other things, the historical record. However, if the record is short then there is the danger that models are based on a time period where events are outside the ‘average’ range resulting in an under- or over-estimation of extreme events from the distributions fitted to observations. In the case of tropical cyclones in Australia, the homogeneous observational record spans less than 40 years, covering only one full phase of the Interdecadal Pacific Oscillation. Here we leverage a much longer (ca. 6,000-yr) chronology of intense paleo-cyclones from five spatially discrete coastal sites between Cairns and Townsville to reassess tropical cyclone wind risk and investigate implications for the insurance industry in this populated area of North Queensland. We inverse-model the physical parameters of paleo-cyclones from surge heights derived from the beach ridge method of Nott (2001, 2003), using a 2D coupled wave and flow model. We then compare the derived frequency/magnitude distribution to the historical data and use the paleo information within a cat loss model to assess the impact on insured losses for this area. The extent to which paleo-tempestology is presently applicable for the insurance industry in Australia, compared to disaster risk management and planning, is also explored.