The lung toxicity of a carbide-cobalt mixture is more important than that of each individual component; the mechanism of this interaction is not understood. The capacity of cobalt metal particles alone and mixed with different carbides to generate hydroxyl radicals was examined with the deoxyribose assay. In a chemical system, cobalt ions and cobalt metal particles (Co) were found to catalyse the degradation of deoxyribose in the presence of hydrogen peroxide. Carbides were able to directly oxidize deoxyribose, but their respective activities did not support such a mechanism to explain the carbide-cobalt interactive toxicity, since there was no direct relationship between deoxyribose degradation ability and cytotoxicity toward macrophages. Tungsten, niobium, titanium and chromium carbides (interactive carbides) were only weak oxidants and conversely molybdenum, vanadium and silicon carbides (non-interactive carbides) were the most potent ones. The ability of cobalt metal to produce hydroxyl radicals in the presence of hydrogen peroxide was not increased by tungsten carbide. The role of reactive radical formation in the toxicity of these particles was further assessed in a macrophage culture model. Catalase (4000 U/ml), superoxide dismutase (300 U/ml), sodium azide (1 mM), sodium benzoate, mannitol, taurine and methionine (all 20 mM) were all unable to protect against the cytotoxic effects of cobalt ions and cobalt metal alone or mixed with tungsten carbide. In conclusion, no evidence was found that production of reactive oxygen species contributes to the elective toxicity of carbide-cobalt mixtures.