One of the challenges to managing fire in many parts of the world is that 'traditional' wildfire risk assessment tools (eg. fire spread modeling software like Behave and FARSITE) often don't work due to the novel fuel types and different fire-climate thresholds. To address this limitation, we use actual fire history to model fire probability as a function of vegetation type, climate variables, human activity and other predictors. This approach is founded on the principles of pyrogeography - the geographic space in which the given combination of vegetation, climate, and ignition sources (the majority of which are people) determine fire occurrence. But the model outcomes are highly applied. We are currently using fire occurrence models in Hawaii to inform watershed protection and restoration planning. There is great potential to apply this type of approach elsewhere where the applicability of more sophisticated fire spread models is limited.