ABSTRACT The US Environmental Protection Agency (USEPA) modified its guidance on determining “method detection limits” (MDL) in 2017. The recommended procedures have not yet been applied to the analyses routinely done on filter samples for USEPA’s Chemical Speciation Network (CSN). This paper applies the new USEPA procedure for estimating MDL to Energy Dispersive X-Ray Fluorescence (EDXRF) analysis of atmospheric aerosol samples collected on filters. The procedure involves estimating MDL by two approaches – statistical distributions of unloaded blank sample measurements and lightly loaded, spiked samples – and sets the MDL as the maximum of these two approaches. Spiked samples at low concentrations were developed using an aerosol deposition chamber to follow this approach. The MDL procedure was initially conducted on one EDXRF instrument, and the spike-based MDL was found to be higher than blank-based MDL for 28 of the 31 elements. The blank-based MDL was higher than the spike-based MDL for Si, K and Fe, which are common contaminants present in filter raw media or arising from EDXRF hardware. The annual verification performed using five EDXRF analyzers demonstrated that the MDL estimated following the USEPA procedure was stable over time and analyzers for all elements except K, which yielded a higher MDL. Implications: Lightly loaded reference materials (RM) were developed for Energy Dispersive X-ray Fluorescence (EDXRF) measurements of elemental concentrations in filter-based particulate matter samples using a novel aerosol generation chamber. These RM were then used to estimate method detection limits (MDL) following U.S. Environmental Protection Agency guidance. These new MDL estimates were compared to alternative estimates and, for most elements, are higher. Our work provides EDXRF users with MDL estimates for each element and an assessment of different MDL estimation approaches.