The standardized mortality ratios for some cancers of the lymphohaematopoietic system were high in an early cohort analysis. Since the presence of large numbers of unexposed workers could conceal risks within a cohort, a case-control study was designed to examine the relationship between estimated exposures and the occurrence of these cancers. The results suggested that the risk for leukaemia was associated with exposure to butadiene and with work in specific areas. Modelling, using rank scores, indicated an increase in the risk for leukaemia with increasing exposure score. Use of cases validated by review of hospital records and selection of a new set of controls did not change the findings. The data indicated that comparison of scores within the same time frame improved the model and increased the estimated odds ratio, suggesting that exposure time or dose rate may prove to be the important variable for risk. Exact measurements from the companies involved showed significant correlations between assigned ranks and level of exposure derived from personal monitoring for butadiene but not for styrene. Thus, use of the measured values might be expected to show an association between leukaemia and exposure to butadiene. The standardized mortality ratio for leukaemia among long-term workers hired before 1960 who had worked in the three plants where the geometric mean butadiene level was three to five times higher than those in the other plants is 1.8 times higher than that of the US population. An appropriate algorithm for comparing cases and controls on the bases of the measured samples is being developed.