When disturbances recur at rates shorter than an ecosystems rate of recovery, it has the potential to result in significant changes to ecosystem structure and function. In western US forests, wildfire activity has increased and many severely burned areas are now re-burning before reforestation occurs. Historically, some of these severely burned patches persisted as montane chaparral in physiographic areas prone to severe fire, but many were early seral communities that transitioned to forest on decadal scales. Subsequent fires responded to the heterogeneity in fuels and maintained a shifting mosaic of different seral stages, promoting resilience across the landscape. This heterogeneity provided a diversity of habitats across space while maintaining forests across time and maintaining many ecosystem services. Past fire suppression and logging practices have homogenized the landscape, creating dense contiguous stands prone to unprecedented patch sizes of high severity. These areas are now re-burning, elevating concerns about transitions to alternative stable states. We hypothesize that repeated severe fire before forest regeneration occurs (< 20 years) will homogenize ecosystems at the landscape scale (spatial patterns of severity across fires). We predict that re-burns will reduce spatial complexity and expand initial high severity patch size due to the high fire intensities generated by chaparral vegetation, and ask: 1) What factors drive re-burn severity? and 2) Does repeated severe fire reduce burn severity heterogeneity? To answer these questions, we are examining three large wildfires that burned in the 1990s and re-burned in the last decade. Mapping fire severity using Landsat imagery, we are examining the drivers of this severity using regression trees to select the most important topographic, fire weather, fire history and vegetation type variables. To examine landscape scale heterogeneity, we are testing differences in established landscape- and patch-based metrics between the initial fires and re-burns. However, since the effects on heterogeneity can vary by spatial scale, a multi-scale approach will be required. We propose adding a field study to examine effects on heterogeneity at the stand scale (vegetation attributes within severity patches), asking: 3) Does repeated severe fire at the stand-scale homogenize vegetation in terms of species diversity, fuels and vegetation structure? We predict that repeated severe fire will homogenize species diversity by narrowing the environmental filter, and that it will homogenize vegetation structure, fuel loadings and fuel continuity by promoting fire-cued shrub dominance and removing regenerating conifer species. We will test these hypotheses by contrasting vegetation data from the following burn histories: twice-burned at high severity in <20 years, severely burned in the initial fire but re-burned at low severity in <20 years, unburned then once-burned at high severity during the re-burn event. We will install 50 plots in each strata, spread over a minimum of three patches per strata, to collect data on species diversity, total cover, cover by functional group and regeneration strategy, overstory live tree and snag density, conifer regeneration density, live fuel continuity and surface fuels. We will use statistical models to test for differences between burn histories, then explore heterogeneity within burn history strata using a variety of dissimilarity measures. Managing for heterogeneous landscapes is an important strategy to promote resilience to the changing climate. Our proposal to quantify the effects of repeated fire on heterogeneity at the stand scale will bolster the relevance of the existing study for managers who work at the unit-level, and will directly inform future management decisions in burned forests.