Prescribed fire is practiced around the world to reduce the effect of unplanned fire, but we hypothesise that its effectiveness is proportional to the mean annual area burnt by unplanned fire, which varies among biomes. Fire history mapping was obtained for six global case studies from a range of biomes: Portugal, Spain (both Mediterranean), Alberta (boreal Canada), Sequoia and Kings Canyon National Parks (montane USA), the Sandy Desert (arid Australia) and Kruger National Park (South African savanna). Leverage is the unit reduction in unplanned fire area resulting from one unit of previous fire as measured at a regional scale over a long period. We calculated leverage for each case study using statistical modelling of annual area burnt, controlling for annual climatic variation. We combined the six leverage values with those from four previously published cases to conduct a global test of our hypothesis. Leverage was high in Portugal (~0.9) and moderate in the Sandy Desert (~0.3). However, the other case studies showed no evidence of leverage: burnt area was not influenced by past fire. In all regions, climatic variation had more influence than past area burnt on annual area burnt. The global analysis revealed a positive relationship between mean area burnt and leverage but only when outlying cases were removed. In biomes with low fire activity, prescribed fire is unlikely to reduce unplanned fire area at all, while for many others, the return for effort is likely to be low. Lessons derived from one biome cannot necessarily be applied to another.