Planned fire is used globally to minimise the risk of unplanned fire, but it is important to measure the return for effort in terms of the reduction of risk per unit area of planned fire. Here, we use 30 years of fire mapping from four subregions of the Sydney region to compare the annual extent of unplanned fire with previous planned and unplanned fire. Using linear mixed modelling, we were able to discriminate the relative influence of previous fire, seasonal rainfall and weather during the peak fire season. The mean annual area burnt over the period was 4.11%, comprising 0.53% planned and 3.58% unplanned. We found that weather during the fire season was the most influential factor. Annual rainfall had a modest negative relationship with unplanned fire area. Past fire had some influence, but the relationship implied that approximately three units of planned fire are required to reduce the unplanned fire area by one unit. Managers would need to burn 5.4% per year to halve unplanned fire extent, a ten-fold increase on recent levels. This would increase the total area burnt, and have other effects that need to be considered (from smoke and greenhouse gas emissions, and changes to biodiversity). © IAWF 2011.