Experimental toxicological studies have offered clear evidence that benzene induces haematopoietic neoplasms, and it is generally accepted that exposure to benzene is a risk factor for leukaemia, in particular for acute non-lymphatic leukaemia. Quantitative aspects of benzene risk assessment are still a matter of controversy, however. In several risk assessments an estimated 50 deaths from leukaemia per 1000 deaths would arise from exposures to benzene of 10 ppm during a working life of 30 years. The assessment presented in this paper leads to lower estimates, which are more in agreement with the weak toxicological data. Furthermore, an approach is presented to incorporate the results of low exposure epidemiological studies into the process of quantitative risk assessment.