This study aimed to estimate exposure-response relationships for mesothelioma and environmental exposure to crocidolite. All 4,659 former residents of Wittenoom, Western Australia (WA) who lived there between 1943 and 1993 for at least 1 mo and were not directly employed in the crocidolite industry, were followed-up through the WA death, cancer and mesothelioma registries, electoral rolls, and telephone books. In 1992, all subjects who should be traced were sent a questionnaire. Exposure levels were estimated from results of periodic environmental surveys and duration of residence. Incidence rates were standardized to the World Population and Cox Regression was used to estimate the effects of exposure on incidence. To the end of 1993, 27 cases of mesothelioma were diagnosed. Mesothelioma cases stayed longer at Wittenoom, had a higher average intensity of exposure, and a higher cumulative exposure to crocidolite than control subjects. The standardized incidence of mesothelioma was 260 per million person-years, and was similar for males and females. The rate increased significantly with time from first exposure, duration of exposure and cumulative exposure. At these levels of crocidolite exposure, there is a significantly increased risk of mesothelioma, which is dose-dependent.