Published epidemiologic studies of butadiene workers, individually and in aggregate, find rates similar to expected for essentially all forms of cancer, with the possible exception of lymphohematopoietic cancers. Reviewers of the literature have disagreed about whether butadiene is a cause of human lymphohematopoietic cancers. This article reviews the available butadiene epidemiologic studies and focuses on the bases for disagreement among reviewers. The disagreement seems to be due to different conventions in applying, explicitly or implicitly, Hill's causal criteria, fostered, in part, by deficiencies in the available epidemiologic studies. Many of these deficiencies will be remedied by updated studies presented at this Symposium or are remediable by other analyses of the existing data.