Owing to the high rates of malignant mesothelioma in workers exposed to crocidolite earlier at Wittenoom and evidence of protection against cancer by vitamin A, a population-based cancer prevention programme providing retinol supplements (25 000 IU/day) was commenced in 1990. The former workers at Wittenoom known to be alive and living in Western Australia in June 1990 constitute the study population. The participants were classified into two groups: those who received supplemental retinol (intervention group) and those who received none (comparison group). The relative rate of mesothelioma for those receiving retinol was estimated using Cox regression, adjusting for cumulative asbestos exposure and age at first exposure to asbestos. Nine hundred and twenty-eight former Wittenoom workers received retinol at some stage of the programme, whereas 1471 workers never received retinol (comparison group). Those who received retinol were younger, had a greater exposure to asbestos and smoked less than the comparison group. There were 65 cases of mesothelioma in the retinol group and 88 in the comparison group. After adjustment, the hazard ratio was 0.99 (95% confidence interval=0.70-1.41). This result did not alter when the participants who received only retinol once or those who received beta-carotene earlier were excluded from the analysis. In conclusion, this study provides little support for possible preventive effects of retinol against mesothelioma in workers exposed to blue asbestos.