Cross-scale spatial and temporal perspectives are important for studying contagious landscape disturbances such as fire, which are controlled by myriad processes operating at different scales. We examine fire regimes in forests of western North America, focusing on how observed patterns of fire frequency change across spatial scales. To quantify changes in fire frequency across spatial scale, we derive the event-area (EA) relationship and the analogous interval-area (IA) relationship using historical and simulated data from low- and high-severity fire regimes. The EA and IA provide multi-scale descriptions of fire regimes, as opposed to standard metrics that may apply only at a single scale. Parameters and properties of scaling functions (intercept, slope, minimum value) are associated statistically with properties of the fire regime, such as mean fire-free intervals and fire size distributions, but are not direct mathematical transformations of them because they also reflect mechanistic drivers of fire that are non-stationary in time and space. Patterns in fire-scaling relations can be used to identify how controls on fire regimes change across spatial and temporal scales. Future research that considers fire as a cross-scale process will be directly applicable to landscape-scale fire management.