Our aim was to describe a vitamin A-based cancer prevention program for former asbestos workers and to check for possible harmful effects by comparing rates of disease and death in study subjects with subjects who chose not to join. All subjects had been occupationally exposed to crocidolite at Wittenoom Gorge between 1943 and 1966; 1,677 subjects indicated interest in the program and 1,203 joined between June 1990 and May 1995. Comparison subjects consisted of 996 former workers known to be alive in Western Australia in 1990 who did not join the program. Program subjects were provided with annual supplies of vitamin A (either synthetic beta-carotene or retinol), help in quitting smoking and dietary advice. The comparison group received only mail contact. Both groups were followed up to December 1994 for vital status and cancer information, and rates of cancer and death from various causes were compared. Mortality in both groups was higher than expected (standardised mortality ratio 1.23 in program subjects and 1.67 in comparison subjects). After adjustment for age, smoking and asbestos exposure, the relative rates in participants compared with non-participants was below I for all examined cancers and causes of death. For mesothelioma and lung cancer, group differences increased with time from entry, whereas other differences dissipated with time. No significant side effects were reported. In conclusion, program participants had significantly lower mortality than non-participants, but the rates of the 2 groups converged with time.