Elemental potassium has been extensively used as an indicator of biomass burning in the source apportionment of PM2.5. We present a method to estimate the fraction of potassium associated with biomass burning (Kb) based on a linear regression with iron that can be applied at any site where PM2.5 chemical speciation is available. The estimated fraction has a significantly greater correlation with levoglucosan (R2=0.63), an organic tracer of biomass burning, than total potassium (R2=0.39). We explore temporal and spatial variability of Kb over a period of six years in the Atlanta area. Kb is larger in spring when biomass burning activity is more prevalent and during weekends due to the use of fireplaces in winter and outdoor charcoal cooking in summer. Kb is the predominate form of potassium for the rural site in this study. The use of Kb in a receptor model results in a lower fraction of PM2.5 apportioned to biomass burning and a greater fraction to mobile sources when compared to the use of total K. Results suggest that Kb is a good indicator of biomass burning.