To provide a scientific basis for determining the health surveillance period of dust-exposed workers, data of a retrospective cohort study was re-analyzed with emphasis on natural course of silicosis. 33640 workers exposed to silica dust who were employed for at least 1 year from 1972 to 1974 in twenty Chinese mines or pottery factories were included as subjects, and were followed up till December 31, 1994. The cohort included subjects from 8 tungsten mines, 4 tin mines and 8 pottery factories. Our results showed that the mean latency of silicosis, for all the cases of the cohorts, was 22.9 +/- 9.8 y. 52.2 % of silicosis was diagnosed approximately 9.1 +/- 5.7 y after the dust exposure had ceased. The progression rates of silicosis from stage I to II and from stage II to III were 48.2 % and 18.5 %, and the duration was 4.1 +/- 0.2 and 6.8 +/- 0.2 y, respectively. The survival times of silicosis stage I , II and III, from the year of diagnosis to death, were 21.5, 15.8 and 6.8 years, respectively. There was 25 % of the silicosis patients whose survival time was beyond 33 y. The mean death age of all silicosis cases was 56.0 y. The death age increased to 65.6 y in the middle of 1990s. Among dust-exposed workers, subjects who became suspected case (0+ ) accounted for 15.0 %. 48.7 % of the suspected silicosis cases developed to silicosis, and the average year from the time of being suspected of the disease to the first stage of silicosis was 5.1 y. The natural characteristics, as mentioned above, varied with different mines and factories. We are led to conclude that silicosis is chronic in nature, but progress quickly. As a serious occupational disease it significantly reduces the life span of exposed workers. The study of its natural history is of importance for the development of health surveillance criteria for dust-exposed workers.