The Pliofilm worker cohort (Rinsky et al., 2002) is comprised of 1,291 benzene-exposed and 554 unexposed workers with follow-up for leukemia and multiple myeloma (MM) mortality through 1996. Five of the exposed workers and three of the unexposed workers died from MM. Prior risk analyses of this benzene cohort were limited to the exposed cohort using U.S. death rates as the external reference population. We propose, for this cohort whose unexposed group is 43% the size of the exposed population (545/1,291 = 43%), that the risk analysis be done by using the unexposed population as an internal reference group. The crude mortality ratio (CMR) for MM for the exposed population (5/1,291 = 0.004) is 0.4 % (95% CI, 0.1-0.9), and CMR for the unexposed population (3/554=0.005) is 0.5% (95% CI, 0.1-1.6). The relative risk is 0.72 (95% CI, 0.17-2.98; Fisher two-tailed pvalue, 0.70). When the analysis is limited to males as all the MM deaths were male, the relative risk is 0.62 (95% CI, 0.15-2.59, Fisher two-tail p-value = 0.45). The CMR analysis indicates that the MM risk in benzene-exposed workers is actually not greater than that of the unexposed. The marginally lower difference seen here is not significant. These results, like others reported in the literature, reinforce the importance of examining an internal reference group analysis when evaluating the potential risks associated with occupational chemical exposures. Analytic use of an internal reference population reduces bias from the healthy worker effect or other risk factors associated with employment.