This study evaluated relations between exposure to 1,3-butadiene (BD), styrene (STY) and dimethyldithiocarbamate (DMDTC) and mortality from leukemia among synthetic rubber industry workers. Subjects were 13130 men employed for at least 1 year during 1943-1991 at any of six plants that manufactured synthetic rubber. Death certificates and medical records identified workers with leukemia. Cumulative exposure estimates were based on plant- and time period-specific process and task characteristics, linked to subjects' work histories. Poisson regression estimated relative rates (RRs) for workers exposed to each agent compared to unexposed workers. Leukemia (N=59) was positively associated with BD ppm-years (RRs of 1.0, 1.2, 2.0 and 3.8, for exposures of 0, >0-<86.3, 86.3-<362.2 and 362.2+ ppm-years; only the RR for the highest exposure category was statistically significant), STY ppm-years (RRs of 1.0, 1.2, 2.3 and 3.2, for exposures of 0, >0-<20.6, 20.6-<60.4 and 60.4+ ppm-years; only the RR for the highest exposure category was statistically significant) and DMDTC mg-years/cm (RRs of 1.0, 2.3, 4.9 and 2.9, for 0, >0-<566.6, 566.6-<1395.1 and 1395.1+ mg-years/cm; the RR for each non-zero exposure category was statistically significant) after adjusting for age and years since hire. After further adjusting each agent-specific set of RRs for the other two agents, a positive but imprecise relation remained for BD and DMDTC but not for STY. The association with BD was stronger for ppm-years due to exposure intensities >100 ppm than for ppm-years due to lower concentrations. BD and DMDTC, but not STY, were positively associated with leukemia in multivariable analyses. The independent effect of each agent was difficult to evaluate because of correlations with other agents and imprecision.