The possibility that excess cancers result from occupational exposures in oil refineries has generated a great deal of interest. Ecological studies and case-control studies in the general population have suggested a positive association between oil industry activity and cancer rates, with more direct evidence provided by studies of refinery employees. The eight investigations of cancer risks among refinery employees are critically reviewed. The methodological strengths and weaknesses of these studies are evaluated with an emphasis on the likely impact on the results. While the results are markedly inconsistent across studies, there is some suggestion of excess risks for melanoma and for brain, stomach, kidney, and pancreatic cancers. Problems with exposure characterization, latency, and potential confounding factors limit all of the studies that were reviewed.