The relative abundance of coke particles, injectant coal-char, and soot found in blast furnace dust provides clues about the operation and efficiency of the blast furnace. The aim of the work has been to determine whether the source of the soots could be related to the injectant coal, in particular, to the nature of the extractable material within it. This paper describes the separation of soot from blast furnace dust and its characterization. Dust samples collected during coal injection tests in a blast furnace and in a pilot scale, "single-tuyere" test rig have been extracted with NMP (1-methyl-2-pyrrolidinone). The parent coals used in these tests were also extracted with NMP; the solvent is highly polar and useful in dissolving various types of soot as well as coal extracts and tars. Fresh coke gave no extract. The coal and dust extracts were characterized by size exclusion chromatography (SEC). Differences were observed between chromatograms of the extracts from the two coals used-the first in the "single-tuyere" test rig, and the second in the blast furnace. SEC was useful in identifying similar features in the chromatograms of extracts from the parent coal and the soot extracted from the blast furnace dust; similar observations were made on the "single-tuyere" test rig dust and the coal injected into it. These findings support the view that tar evolution from the injectant coal contributes directly to soot formation and is a major contributor to levels of dust emission observed at high coal injection rates.