Seed shipments, silos and storage houses often contain weed seeds or seeds of restricted crops such as undeclared genetically modified (GM) varieties. Random sub-sampling is the favoured approach to detect unwanted biological materials in seed lots but is prohibitively expensive or else ineffective for the huge volumes of seeds moved in commercial operations. This study uses maize and cowpea seed admixtures as an exemplar to evaluate the feasibility of using aerosol sampling of "seed dust" as an alternative to seed sub-sampling. In an initial calibration phase, qPCR of the rbcL barcode followed by high-resolution melting (HRM) of a DNA titration series revealed a strong linear relationship between mix composition and HRM profiles. However, the relationship became skewed when flour mixes were used to build the titration, implying a DNA extraction bias favouring cowpea. Aerosol samples of seed dust above a titration of mixed seed samples were then collected along vertical and lateral axes. Aerosols were characterised by light microscopy, qPCR-HRM and next-generation DNA sequencing (Illumina MiSeq). Both molecular approaches again showed bias but this time in a reverse direction to flour samples. Microscopic examination of the aerosol sample suggested this divergence could be attributed to differences in abundance of airborne starch particles. Despite the bias, it was nevertheless possible to estimate relative abundance of each species using the abundance of minibarcodes. In light of these results we explore the feasibility of aerosol sampling for commercial seed lot characterisation.