A previous retrospective follow-up mortality study of white men employed at an oil refinery found an excess of leukemia deaths, especially of acute myeloid leukemia (AML). A casecontrol studv was done to determine if the excess could be related to any particular jobs or departments at the refinery or to benzene exposures. The study includes 14 cases with leukemia (eight cases with AML) and 60 matched controls. No particular job or department was identified as the likely source of the excess leukemia deaths. Furthermore, the cases apparently had not been exposed to more benzene than had the controls. Thus, the reason for the excess leukemia at this refinery was not identified. The limitations of the study are discussed.