Former blue asbestos workers known to be at high risk of asbestos-related diseases, particularly malignant mesothelioma and lung cancer, were enrolled in a chemo-prevention program using vitamin A. Our aims were to compare rates of disease and death in subjects randomly assigned to beta-carotene or retinol. Subjects were assigned randomly to take 30 mg/day beta-carotene (512 subjects) or 25,000 IU/day retinol (512 subjects) and followed up through death and cancer registries from the start of the study in June 1990 till May 1995. Comparison between groups was by Cox regression in both intention-to-treat analyses and efficacy analyses based on treatment actually taken. Median follow-up time was 232 weeks. Four cases of lung cancer and 3 cases of mesothelioma were observed in subjects randomised to retinol and 6 cases of lung cancer and 12 cases of mesothelioma in subjects randomised to beta-carotene. The relative rate of mesothelioma (the most common single cause of death in our study) for those on retinol compared with those on beta-carotene was 0.24 (95% CI 0.07-0.86). In the retinol group, there was also a significantly lower rate for death from all causes but a higher rate of ischaemic heart disease mortality. Similar results were found with efficacy analyses. Our results confirm other findings of a lack of any benefit from administration of large doses of synthetic beta-carotene. The finding of significantly lower rates of mesothelioma among subjects assigned to retinol requires further investigation.