Literature data of epidemiological studies, carcinogenicity studies and in vitro studies on inorganic substances were surveyed with the aim to determine sensitivity and specificity of in vitro tests of nanomaterials. Asbestos, quartz and chromium and cadmium compounds were assigned to classes of highest carcinogenic potency. After 20 years of occupational exposure to long-term average concentrations of 0.5mg/m(3) of these dusts - or to even lower concentrations - an epidemiologically detectable increased lung cancer risk has to be expected. In contrast, diesel engine emissions, some nickel species and "ultrafine" versions (nanomaterials) of titanium dioxide and carbon black were also carcinogenic in inhalation studies, but show varied epidemiological results. The high frequency of lung cancer in the male general population due to cigarette smoking hampers unequivocal detection of occupationally caused lung cancer risks. Based on the experience from the inhalation studies, workers had to be exposed to long-term concentrations of 1mg/m(3) or more to identify epidemiologically a clear cause-and-effect relationship for a specific substance of intermediate potency. Respirable granular biodurable particles without known significant specific toxicity with primary particle sizes of more than 1μm have also shown carcinogenicity in rats. Their potency was even lower; and partially results after instillation rather than inhalation are available. Nearly all types of nanomaterials and control dusts used in the in vitro assays showed genotoxic effects in cell cultures (e.g., CoCr particles, diesel soot, SiO(2) crystalline and amorphous, TiO(2), carbon black), but not consistently in all studies; overall, the proportion of positive results was about 50%. No clear correlation of the probability of a positive in vitro test with particle properties was seen. I recommend trying and calibrating a sensitive in vitro model (e.g., micronucleus assay) against the described rank order of carcinogenic potency by testing a series of inorganic substances.