Introduction: While considered extreme events, wildfires will lengthen and strengthen in a changing climate, becoming an omnipresent climate-sensitive exposure. However, few studies consider long-term exposure to wildfire fine particulate matter (PM2.5). Here, we present a conceptual model to assess long-term wildfire PM2.5 exposure and evaluate disproportionate exposures among marginalized communities. Methods: We used 2006-2020 California census tract-level daily wildfire PM2.5 concentrations generated from monitoring data and statistical techniques to derive five long-term wildfire PM2.5 measures. We classified tracts based on their CalEnviroScreen (CES) score, a composite measure of environmental and social vulnerability burdens, and their racial/ethnic composition. We determined associations of (a) CES score and (b) racial/ethnic composition with the five wildfire PM2.5 measures using separate mixed-effects models accounting for year and population density. To assess differences by year, models included CES or race/ethnicityyear interaction terms. Results: We conceptualized and compared five annual wildfire PM2.5 exposure measures to characterize intermittent and extreme exposure over long-term periods: (1) weeks with wildfire PM2.5 >5μg/m3; (2) days with non-zero wildfire PM2.5; (3) mean wildfire PM2.5 during peak exposure week; (4) smoke-waves (2 consecutive days with 25μg/m3 wildfire PM2.5­); (5) annual mean wildfire PM2.5 concentration. Within individual years, we observed exposure disparities, but generally did not when averaging over the study period. Non-Hispanic American Indian and Alaska Native populations, however, were consistently over-represented among the exposed population compared to their California-wide representation. Conclusion: We found that wildfire PM2.5, measured via five metrics, disproportionately affected persistently marginalized California communities—with substantial year-to-year variability.