Summary. To date, all risk assessment studies on benzene have been based almost exclusively on epiderniological data. Wehave attempted a more integrated and quantitative evaluation of carcinogenic risk for hurnans, trying to utilize, in addition to the epidemiological data, all data available, specifically data on metabolism, genotoxicity, and carcinogenicity in small rodents. An integrated evaluation of the globality of the available data seems to suggest a progressive saturation of metabolic capacity both for man and rodents between 10 and 100 ppm. The most susceptible target cells seem tobe different in humans (predominant induction of myelogenous leukemia) and small rodents (induction of a wide variety of tumors). Nevertheless, both epidemiological and experimental carcinogenicity data tend to indicate a flattening ofthe response for the highest dosages, again suggesting a general Saturation of mechanisms of metabolic activation, extended to different target tissues. From a quantitative point of view, the data suggest a carcinogenic potency at 10 ppm two to three times higher than that computable by a linear extrapolation from data in the 100 ppm range. These observations are in accord with the recent proposal of the European Economic Community of reducing benzene time-weighted average occupationallevels from 10 to 5 ppm.