An excess of lympho-haematopoietic (LH) cancers has been identified in the Australian petroleum industry through the Health Watch surveillance programme. A nested case-control study is being conducted to investigate this excess. This paper describes the methods used to provide quantitative estimates of benzene exposure for each of the subjects in the case-control study. Job histories were compiled for each subject from interviews and company employment records. Site visits and telephone interviews were used to identify the tasks included in each job title. Details about the tasks such as their frequency, the technology in use and about changes that had taken place over the years were also gathered. Exposure dated back to the late 1940s for a few subjects. Collaborating petroleum companies provided recent benzene exposure monitoring data. These were used to generate Base Estimates of exposure for each task, augmented with data from the literature where necessary. Past exposures were estimated from the Base Estimates by means of an exposure algorithm. The modifying effects of technological changes and changes to the product were used in the algorithm. The algorithm was then computed to give, for each job, for each subject, an estimate of average benzene exposure in ppm in the workplace atmosphere (Workplace Estimate). This value was multiplied by the years for which the job was held and these values summed to give an estimate of Cumulative Estimate of benzene in ppm-years. The occupational hygienists performing the exposure assessment did so without knowledge of the case or control status of subjects. Overall exposures to benzene in the Australian petroleum industry were low, and virtually all activities and jobs were below a time-weighted average of 5 ppm. Exposures in terminals were generally higher than at refineries. Exposures in upstream areas were extremely low. Estimates of Cumulative Estimate to benzene ranged from 0.005 to 50.9 ppm-years.